Six deploy and helper workloads are restricted; the count of workloads
without a narrow rule went from fifteen to nine. The block is proven from
inside the pod - three truly external targets refused, CoreDNS answering as
the positive control - and the pod is distroless, so the probe ran in an
ephemeral container sharing its network namespace and labels. That is the
method the remaining waves need and it is written down rather than
rediscovered.
Rehearsing the rollback corrected a claim from Gate 2. "Flux makes a git
revert a rollback of minutes" holds only with an addition. The revert took
about forty seconds; restoring afterwards did nothing for four minutes,
although git.lab had the commit and the mirror reported finished. Only a
forced refetch of the GitRepository moved it.
So a rollback is a revert plus a nudge to the source. Someone who only
reverts and waits sees nothing for up to a poll interval and may conclude
the rollback itself is broken while it is merely slow. That belongs in the
record before a wave touches something users notice.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
The owner pointed out that rohana is still reachable over the private
network and need not be called externally. Measured and correct: 10.0.0.3
answers as Gitea for the public host header and serves a valid Let's
Encrypt certificate for that very name, validated with full verification.
So the private path is not just reachable but TLS-clean, and no service
needs reconfiguring. What is missing is only the internal pointer - the
Corefile already imports the custom override directory, the ConfigMap
simply does not exist yet.
Two of the four class B workloads therefore stop leaving the cluster, and
the Gitea exception recorded in AGENTS.md becomes an internal rule. The
storage box stays external: no private path was named for it and I am not
assuming one.
The price is written into the design rather than skipped. An internal
pointer turns two paths into one - if 10.0.0.3 is down, rohana is
unreachable from the cluster although the public route would work, and the
failure would look like "Gitea is gone" instead of "the private path is
gone".
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Two findings from the preparation change the design.
There is no split-horizon DNS: every public name of the platform resolves
inside the cluster to the node's external address. Services that reach each
other by public name therefore leave the cluster and come back through
Traefik.
And traffic to the node's own address is not subject to the policy.
Measured twice, from two workloads restricted since the 20th, with
different tools: the storage box, rohana and 1.1.1.1 are all blocked, the
node address is reachable, and CoreDNS answers as the positive control. The
restriction works - just not against the node itself. That is a property of
the k3s enforcement, not a manifest error, and no NetworkPolicy can close
it, because policies allow rather than forbid.
It cuts both ways. Class A grows and gets safer, because services talking
over public names keep working under restriction. And the ten workloads
restricted yesterday can still reach anything published through Traefik -
the gap between what the manifest promises and what holds belongs written
down rather than smoothed over.
I also walked into the trap my own acceptance criterion warns about. The
first counter-probe used /dev/tcp in a container whose sh does not have it,
so everything came back "blocked", including the reachable node. Same
mistake as on the 20th, in the undertaking whose criterion 4 names it.
Repeated with nc and a positive control; only then was the result worth
anything.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Two constraints measured rather than assumed. k3s has no separate CNI or
policy pod, so enforcement is standard NetworkPolicy: CIDRs only, no DNS
names. And the cluster is single-stack IPv4 - a v6 connection attempt fails
with OSError - which matters more than it looks: had it spoken IPv6, every
IPv4-only rule would have been a gate that looks tidy and holds nothing.
The fifteen workloads fall into three classes: internal only, exactly one
pinnable target, and genuinely broad. The manifests do not yield the
targets - external addresses live in SOPS secrets and in the images - so
class B comes from resolved names instead.
ClamAV is the hard case and is not talked away. Its signature source
resolves to Cloudflare with rotating addresses. Pinning them breaks
silently at the next rotation - signatures age, the service keeps running,
nobody notices - which is the failure class this project built half its
checks against. Allowing Cloudflare's ranges would look like a restriction
and barely be one. So its egress stays, with the reason in the manifest,
and mirroring signatures internally is named as the later option.
Rolled out in three waves by risk rather than one commit, so a failure says
which service stumbled. The counter-proof will use Python from inside a
pod against a target that is demonstrably reachable with broad egress -
on the 20th both a service IP and /dev/tcp in a dash container produced a
false "blocked".
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Assigned by sorb. Size L: fifteen workloads, a real decision per service,
and mistakes land in production.
The acceptance criteria are counted in the running cluster rather than read
off the manifests - that distinction is what produced this undertaking in
the first place, since the issue's own remaining list was a week out of
date. Zero workloads without an egress rule, every exception naming where
and why with evidence that the service fails without it, a pre-agreed
smoke list, and a demonstration from inside a pod that a forbidden target
actually fails.
Non-goals name the temptations: no ingress changes, no new model or tool,
and no switching a feature off to save a rule. Synapse needs broad egress
for URL previews and already has the right control in
url_preview_ip_range_blacklist; disabling the feature would be the wrong
lever, as the issue has said since the 20th. flux-system is excluded too -
it pulls from the network and is the thing that rolls this change out, so
cutting its ground is a separate act with its own fallback.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
All three rules evaluate healthy after the pull and reload, with the prefix
matcher active and the probe named in the message.
Healthy only means a rule can be evaluated, not that it sees anything, so
the series underneath were checked as well. The new cronjob has info and
created series but no last_schedule_time - a manual run does not set it and
the first scheduled one is 04.09. Without the "or kube_cronjob_created"
fallback the rule would stand on an expression with no series for this
probe and could never fire: green, healthy and blind.
Whoever wrote that fallback a week ago did it as a precaution. Today it
carries real weight for the first time, on a cronjob that did not exist
when it was written.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
restore-drill-media, on the 4th at 05:20, an hour after the database
probe. The name is the alerting: BackupJobFailed already matches
restore-drill.*, so a failure needs no new rule and reaches the
maintenance room through the single alertmanager route.
Every link is measured rather than assumed. A control run with a
deliberately damaged file reported one mismatch and failed the job. The
job name was checked against the rule's regex. The route was read from the
config. And delivery works today: seven notifications sent, none failed,
alertmanager up.
The probe re-proves its own comparison on every run: after passing, it
alters one shared file by a byte and fails with "this probe proves
nothing" if the comparison stays quiet. That a comparison has only ever
said "equal" is a guess, and running monthly does not change it.
Silence is covered too - the stale and missing alarms now match the prefix
and name which probe is affected - but those two rule changes sit in
threadnet-operating and are not live: the operating stack does not pull by
itself, and Prometheus still serves the old expressions, measured through
its rules API. The failure alarm is unaffected and already armed. Written
down as open rather than reported as done.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
The point named when #0030 was closed is done. A job in the cluster
extracted the newest archive into an emptyDir and compared it against the
production PVC mounted read-only: every one of the 309 files present in
both is byte-identical by sha256.
That is a stronger statement than the database drill makes. There it was
shown that rows arrive; here that the restored files are the running ones.
Three flaws in the test, all mine, and all written down. busybox find has
no -printf, so the byte counts came out as a silent zero beside real
numbers. The first run compared totals and counted every cache prune as a
mismatch, when the backup is a snapshot and production keeps running - the
intersection is what proves anything. And my exception knew url_cache but
not url_cache_thumbnails, so it reported a finding outside the cache that
was not one; the filter worked, its list did not.
Still unproven: the emergency path itself, copying back into the PVC. What
is proven is that the data come out of the archive complete and unaltered.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
The subject of the issue was whether the backups were a guess. That was
settled on 14.08. with restored row counts rather than an assurance, and
the title has been wrong ever since. It also sat at in-progress for a week
without work. Third issue today with the same stale-head class.
Two things measured while closing, neither of them in the text:
restore-drill has never fired on schedule. lastScheduleTime is none - the
job was created on 14.08. and runs on the 4th, so the first automatic run
is 04.09. Only the manual run has passed. Every other cronjob in the
cluster shows a fresh timestamp; this one does not. Not a fault, but
"automated" and "proven" are not the same thing.
The Synapse media are backed up and the restore is untested. synapse-backup
mounts the PVC and archives /media/media_store - the run of 21.08. held 320
files and 216 MB, which would not add up without media. The way back is
written down and explicitly not automated, and nobody has walked it.
The homelab item is dropped rather than carried anywhere: the homelab is
not part of this project, and it should not have been listed as remaining
work in the first place.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Both issues described work that no longer exists. Without measuring, two
tasks would have entered planning half-done.
0083 is done. No single-file mount remains in the compose file; the running
exporter emits both metrics that only exist since the 19.08. change; and
the timing closes the chain - the mount commit carries 12:00Z while
Prometheus and Alertmanager have been up since 16:34Z and Loki since 17:08Z
the same day, so the redeploy came after the change. A "yes, it was rolled
out" would have claimed the same and proven nothing.
0088 stays, with a title that is no longer wrong. It said "13 ingress
rules, 1 egress", which was the state on 06.08.; measured today there are
ten workloads restricted to internal traffic and fifteen still open, not
the eleven the issue listed.
Measuring found more than it corrected. wikijs itself is unrestricted while
its database is, which nobody intended and nothing recorded. And the
namespaces authentik and monitoring carry only the metadata block, so their
egress is entirely open - that is not in the issue at all.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Size S owes no gates, but the ladder section is mandatory even when nothing
was built - an absent record and an ignored rule look identical.
"One ledger per session" does not quite hold here. This session's ledger
was closed at Gate 5 and then this task arrived. A second ledger
contradicts the wording; no ledger contradicts the sentence above it, that
every rule owes a trace or it is decoration. I chose the trace and wrote
the contradiction down rather than resolving it quietly: a session can hold
more than one finished run, and the concept currently knows only sessions.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Three task briefs lived only in a chat and in a home directory. They are
the only record of what was actually handed over, and therefore the only
basis on which a result can later be judged to match its request.
They land beside textbloecke.md rather than in a new directory: that page
already holds "short, copyable blocks you put in front of a session", and
these are the same idea at full length. Each carries its state at the top,
because a brief that reads as pending when it is done is the stale-head
failure this project keeps finding.
The third one exists for a reason worth stating: an agent judging its own
run justifies rather than checks, so that verdict must not be written by
the session that produced the run. And such a brief carries no findings of
the author's own - handing them over buys a confirmation instead of a
check, and then the separation is only cost.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Closing the run turned status to closed, and the judge - which checks a
closed ledger strictly and an open one loosely - reported fifteen findings
where it had reported none. All mine, in three classes: no ladder entry
carried the commit it became effective in, gate 4.4 carried no approval,
and the slices stood as gate rows although the template expects them in the
design document at size L ("size M records its slices").
The slice rows are removed, which touches existing rows that "appended,
never rewritten" protects. They should never have been there; the design
document has carried them with their commits all along.
Worth keeping, because it is the whole argument for the ledger: while the
run was open the record looked complete and was not. Only the act of
declaring it finished made the check strict enough to say otherwise.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Five statements this undertaking contradicts, named rather than left
standing. Two of them are in this document's own earlier gates: the Gate-2
mapping had FB-06 as open when the framework had split the item and closed
that half in v0.1.3, and had FB-09 and FB-11 as open when v0.3.0 covers
them partly. Both only surfaced because the states were checked against the
rule text of the tags instead of the issue status.
The third is FB-12's "point 3 is fixed", whose reasoning - the error is
forgetting, not doing it wrong - was falsified by the same task. That page
is revised rather than appended to, which is the practice this project
harvested and the framework shipped in v0.1.3.
Criterion 5 stays partly met. The deterministic half ran against both of
the day's ledgers without findings; the verdict artifact must not be
written here, because its own template holds that a verdict produced in the
working session is void whatever it says.
Moving the document one level deeper broke three relative links, which
validate.py caught before the commit. Same class as everything else today:
the tool saw what the plan did not.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
The row was written in the same commit as the work, so it could not name
that commit's sha - the second time in this run. The parallel run shows the
order that works: the work commit first, then a commit for the ledger row.
That is what "rows are appended as gates close" means.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Slice 4.4. judge.py caught gate 4.3 naming no commit - the row was written
before the commit existed and the sha never filled in. Classified as model
failure, correctly: the rule is clear, the template says so, I did not do
it. Filled in, then clean. A fabricated sha is refused and fails closed.
Both ledgers of the day pass the deterministic half, including the parallel
run's, which this session did not write.
The verdict artifact is missing on purpose. Its own template says a verdict
produced in the working session is void whatever it says, and an agent
judging its own run justifies rather than checks. Acceptance criterion 5 is
therefore partly met, recorded as partly rather than ticked.
First rule-coverage measurement: 14 of 23 rules observable, nine not -
among them "every line traces to the request", "a check owes proof it can
fail" and "incidents get an AAR". Decoration until they carry a duty to
leave a trace. That is the claim this whole series started from, now a
number instead of a suspicion.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Slice 4.3. Every finding is now a stolpersteine page carrying its state,
and the mapping is discharged mechanically: twelve findings in the
register's last hand-maintained revision, twelve pages naming their origin,
zero unassigned.
The states were verified against the rule text of the tags, not derived
from issue status - and that turned up two errors in my own Gate-2 mapping.
FB-06 was listed as open; the framework had actually split the combined
item and closed that half as issue 0028 in v0.1.3, whose WORKFLOW.md
carries "Name what this work made false" verbatim. FB-09 and FB-11 were
listed as open too; v0.3.0 covers them partly through the ledger and the
ladder's trace duty, so they are `partly` with the version named.
Two are `declined`: the framework considered them and will not cover them,
so they remain ours. That state exists because writing `open` for a decided
matter is the failure class this undertaking exists to clear.
The texts are not edited. They came out of git at the register's last
revision and moved unchanged; only repo-relative links were rewritten,
because inline links resolve against the file.
Also done: the duplicated half of the ADR rule is gone from the project
section - permanent exceptions have been upstream since v0.1.2 - and three
descriptions that still called the register an inbox now describe the
pages, the generated overview and the signpost.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Slice 4.2. The receipt duty from ADR-0024 was built yesterday on the
premise that "the error is forgetting, not doing it wrong". The same
upgrade falsified that on the same day: the receipt for schema.yaml was
written, it is truthful, it names correctly what was carried over - and two
fields were lost anyway. Nobody forgot anything.
pruefe_upstream_drift.py now holds our schema.yaml structurally against the
baseline: every type, field, enum value and required entry the baseline
carries must exist here. Adding is allowed, losing is not. Checked
counterfactually against the real loss, which it reports.
For the two other extended files there is no equivalent - they are Python,
and no structural comparison exists. That half stays open and is written
down rather than glossed over.
Building it taught the check two things about itself. It broke the existing
selftest because that fixture wrote schema.yaml as prose - the same
unrealistic fixture FB-12 already records as having made an assertion
worthless for this very script. And my counter-control reported "nothing"
for the removed crash guard, because a crash returns the same exit code as
a failure; only checking for a traceback showed the script never reaches
its summary without it. Both times the control was blunt, not the check.
Sixteen assertions, six deliberate breaks, none uncovered.
The occasion is recorded for the next harvest as two stolpersteine pages -
the first use of the mechanism ADR-0009 decided: a receipt certifies
attention rather than completeness, and a file cannot be authoritative and
immutable at once.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Slice 1, the tracer bullet: schema, generator and one migrated page, so the
whole chain runs before eleven more depend on it.
Two fields were missing entirely. The v0.3.1 merge did not carry over
wiki-page's status and harvested_in, so the mechanism ADR-0009 decides was
not actually available here. A field-by-field comparison against the
vendored baseline found exactly those two and nothing else - the gap the
previous run predicted when it noted that reconciling the extended files is
a manual step with nothing to contradict it.
The generator gains a clustered section and writes the signpost; collect()
and apply_rules() already existed, so the change is a filter and a fifth
rule rather than a second reader or a new checker.
Running it corrected one of my own design errors immediately. The "without
state" group was written as a warning, and it flagged two perfectly correct
pages: status is optional per ADR-0009 and only meaningful on a page that
tracks a pattern. A permanent complaint with no subject trains people to
ignore the section, so the group is now neutral - with the downside written
into the code, since a pattern that lost its state now looks like an
ordinary page.
Eight controls, seven of them deliberate breaks: both generated files go
stale on a hand edit, an invented state value is refused, the new value is
accepted and clusters correctly, and harvested or declined without a named
version now fails.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
The owner objected to the plan carrying two findings as "open" when the
framework has considered and declined them, on the grounds that the enum
has no better value. He is right: replacing a state the field cannot
express with a state that is wrong is the exact class this undertaking
exists to clear.
Extending is established practice here rather than an exception - the
extension block in schema.yaml already declares two enum extensions, and
ADR-0024 covers the file. The portability cost is near zero because these
pages never travel upstream: ADR-0008 sends generalized failure classes,
which carry no frontmatter.
The value is `declined`, not `rejected`. Our issue enum already uses
`rejected`, and matching vocabulary would be cheaper, but the meanings
diverge: a rejected issue is finished, while a declined pattern persists
and stays ours to live with. Anyone reading `rejected` would think the page
is disposable.
Measured while deciding where to document it: HERKUNFT.md cannot be
maintained at all. It sits under docs/sources/ and the lock check refuses a
test commit against it, while ADR-0024 declares its table authoritative and
requires it to stay in step with the pair list. A file cannot be both the
maintained truth and immutable. Recorded for the next harvest; an accepted
record is not repaired in passing.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Option A said FRAMEWORK-BEFUNDE.md would cease to exist as a file. It
cannot: ADR-0024 is accepted, links it twice, and an accepted record is
never edited. Deleting the file breaks validate.py at a line nobody is
allowed to repair.
Resolved without touching the record: the file stays as a generated
signpost of a few lines carrying no inventory of its own. The data live in
the pages, the overview in STATUS.md, and the file only says where both
are. Every reference stays valid and nothing can rot, because a signpost
holds nothing that could. It also means AGENTS.md needs exactly one change
- the duplicated rule - rather than two.
Eleven new pages, not twelve: FB-02 already owns one. collect() already
reads artifacts by directory and type, so the generator gains a filter
rather than a second reader.
The assertions that matter are the ones about absence: a page without a
state must show up as a finding rather than vanish, and harvested without
harvested_in must fail. The difference between "no state set" and "not
there" is the failure class this whole register was started for.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
The owner's condition on approving Gate 1 was to check whether v0.3.1 is
already adopted before doing anything. It is, as far as the rules go: the
upstream part of AGENTS.md and all of WORKFLOW.md are byte-identical with
the vendored baseline, every new section is present, and the drift check
passes. What is not done is the content: no stolpersteine page carries a
harvest state, nothing generates the findings overview - neither our
gen_status.py nor the upstream one knows the concept - and judge.py is
wired nowhere.
Two ladder results worth keeping. The overview extends gen_status.py rather
than becoming a third generator. And FB-02 already owns its page, linked
from the register twice, so it gets a state instead of a duplicate.
The harvest state is now derivable instead of asserted: the upstream issues
carry it. Three of ours are done there, six open, and two were rejected
with reasons.
That last group exposes a gap in the state space. A pattern the framework
considered and turned down is neither open nor harvested, and the enum has
no third value. Carrying it as open asserts a process that no longer
exists - the very class FB-04 describes. Recorded as a finding for the next
harvest, not invented around here.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
The undertaking was set up as the whole two-stage adoption and had to be
re-cut on measurement: a parallel run had already completed stage 1 the
same day and recorded it in its ledger. Gate 1 is therefore rewritten
rather than extended - a plan that announces work already done is the same
defect as an issue whose head asserts a diagnosis its appendices refuted.
What is left is the part that fails quietly if the adoption is treated as
finished. The twelve findings still live in a hand-maintained register,
while the framework's ADR-0009 gives them a home with a harvest state.
Four framework versions have shipped since they were handed over, so
"a released version covers this" is measurable for the first time - and
unmeasured. And judge.py is adopted but wired nowhere, with no verdict in
existence.
One decision from 2026-08-20 was deliberately reversed by that parallel
run, with approval and a ladder entry: upstream's check_locked.py was not
adopted because ours does the same job and only had a defect, which was
fixed. Not re-litigated here.
This run keeps a ledger, as now required. It records both the re-cut and a
working-tree contamination during the survey.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
FB-12 point 3: the declared-extended files are deliberately outside the
byte comparison, so nothing reported when upstream changed them. Two of
three were touched between v0.1.1 and v0.3.1 and it only surfaced because
someone thought of it.
Two checks added to the script that already owns the subject, rather than
a second script beside it:
* completeness — every vendored file belongs to exactly one class.
check_harvest.py and judge.py were adopted byte-for-byte during the
upgrade and never entered the pair list; nothing compared them.
* reconciliation — if upstream touched an extended file between the
previous and the current baseline, ABGLEICH.tsv of the new baseline
must acknowledge it.
The acknowledgement lives in its own file on purpose. The first draft
searched HERKUNFT.md's prose for the filename, which is always there in
the inventory table: strict-looking and inert. Its selftest missed that
because the fixture was an unrealistically empty note; the counterfactual
against real history caught it.
Ten assertions, twelve deliberate breaks, none uncovered. Two of the
breaks exposed two useless assertions.
AGENTS.md requires an ADR for every permanent exception: documenting one
instead of deciding it is an error. schema.yaml, validate.py and
gen_status.py have diverged from the vendored baseline since the
migration on 2026-08-11, described in two places and decided in none.
FB-12 named the gap during the v0.3.1 upgrade.
A prefix comparison was measured and rejected: unlike AGENTS.md, these
extensions are interleaved — component sits between issue and wiki-page,
the two rules sit inside apply_rules, and gen_status replaces the issue
table rather than appending to it.
The honest price is in the consequences: the reconciliation on upgrade
has no machine contradictor. Measured on this very case — upstream
touched two of the three between v0.1.1 and v0.3.1 and nothing reported
it.
The baseline moved v0.1.1 to v0.3.1 on 2026-08-21, the first real run of
a procedure that had been considered settled since v0.1.2. It failed at
three places: WORKFLOW.md could not be copied byte-for-byte because it
links to the framework's own decision records; pruefe_sperrliste forbade
the upgrade itself, unable to tell an addition under docs/sources from a
change; and the three declared-extended files had to be reconciled by
hand because nothing reports that upstream touched them.
One root cause: the procedure was described and never executed. Two of
the three are fixed, the third is open — which is FB-11 on a new subject.
Our own positive-control rule, not applied: the adoption question was
closed on the description of a procedure. Nobody ran it.
pruefe_sperrliste read --name-only, so it could not tell a modified
source from a newly added one and flagged both. That made a framework
upgrade impossible by construction: an upgrade must place a new baseline
under docs/sources/upstream/, and the lock rejected exactly that.
Upstream's check_locked.py draws the same line and asserts it ('adding a
new source is not an edit'). Modification and deletion stay locked.
Three minor releases at once. Byte-identical surface: only WORKFLOW.md
actually changed (+128/-2); CLAUDE.md and the four templates are
untouched. AGENTS.md takes the new upstream prefix (+22/-2) and keeps our
project section unchanged.
The two declared-extended files were reconciled by hand, because nothing
compares them: validate.py gains upstream's check_vendored_portable, and
schema.yaml gains vendored, project_section_marker, the judged link field
and the ledger and verdict types. gen_status.py needed nothing — upstream
did not touch it. Recorded in the new HERKUNFT.md, including that this
reconciliation has no contradictor and will be forgotten next time.
Newly adopted: check_harvest.py and judge.py, plus docs/ledger/ and
docs/verdict/ with their templates. Deliberately not adopted:
check_locked.py — pruefe_sperrliste.py has done that job here since
2026-08-20, and two tools for one rule is maintenance without gain.
The v0.1.1 baseline stays where it is; docs/sources is immutable.
The list of names that must never leave in a harvest lives here rather
than in the framework: the framework must not store the names it exists to
keep out. That placement is the framework's own decision (its ADR-0008),
and this file is our side of it.
Two rules carry the reasoning, both learned expensively. A term the
framework itself uses is shared vocabulary and belongs off the list -
listing it produces noise, and noise teaches people to skip the check. And
a green run proves nothing about absence: a denylist finds only what
somebody thought of, which is exactly the confusion behind the near-leak
of 2026-08-13.
All eleven findings were handed over on 2026-08-20 as nine issues of the
next framework iteration. They stay open here regardless: handed over is
not harvested, and by this file's own rule a finding only counts once the
next iteration actually covers it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Size class and closing status are carried from now on. That settles the
behavioural half only: it was already the rule before, so what changed is
attention, not mechanism. FB-01 shows how long attention alone holds, so
the finding stays open.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
sorb asked why the ponytail ladder (AGENTS.md section 1, lines 26-29) is
never used. It is not: it was never named and never demonstrably climbed.
Measuring the whole rule set shows ponytail is only where the pattern
surfaced. Every AGENTS.md rule backed by a script was kept without
exception; every rule without one was broken or ignored:
- the four mandatory completion statuses appear only in the rule files
themselves and their baseline copies, in no work product
- size classes S/M/L likewise, only in rule text
- WORKFLOW.md demands a design doc for L; docs/design holds exactly one,
from the migration itself, while the upstream merge, v0.6.0, the
typecheck gate, the egress lockdown and the game host onboarding ran
without one
The unguarded rules all sit in the byte-identical baseline, so this repo
cannot add enforcement without breaking pruefe_upstream_drift.py. That is
why this belongs in the harvest list rather than in a new script - and a
script could only count rituals, not judgement (FB-03).
FB-01 suspected the mechanism at one instance and now points here.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F2Q4Ri8NGwyTZzScvKnWFM
Entscheidung sorb. Drei Ergaenzungen, alle NACH der Baseline-Marke - die
uebernommene neckbeard-Fassung bleibt byte-treu, pruefe_upstream_drift meldet
0 Fehler:
- Werkzeug-Tabelle um pruefe_sperrliste.py
- neuer Abschnitt 'Nach Deploys und Vorfaellen' mit der AAR-Pflicht
- Verweis auf FRAMEWORK-BEFUNDE.md als Eingangsliste der Framework-Ernte
Die AAR-Pflicht stand bisher nur in roadmap.md. Auf AGENTS.md verweist aber
jedes CLAUDE.md - wer der Zeigerkette folgte, fand sie nicht. Am 2026-08-20
fehlten dadurch sechs AARs auf einmal.
⚠️ Im Text steht ausdruecklich, dass kein Werkzeug diese Pflicht erzwingt.
FB-01 ist damit nur zur Haelfte behoben: die Wissensluecke ist zu, die
Durchsetzungsluecke nicht. Das offen zu benennen ist ehrlicher, als sie mit
einem Zeiger fuer erledigt zu erklaeren.
Umsetzung von FRAMEWORK-BEFUNDE FB-10. Anlass: Commit 6652125 hat die
angenommene ADR-0022 veraendert, obwohl AGENTS.md sagt 'binding; never edited,
only superseded'. Der Verstoss wurde gepusht und erst durch Zufall bemerkt.
STATUS.md traegt dasselbe Verbot und wurde nie verletzt - weil gen_status
--check widerspricht. Dieses Skript ist der fehlende Widerspruch.
Erlaubt bleibt genau die Aenderung, die die Vorlage vorschreibt: superseded_by
und status in der abgeloesten ADR setzen. Jede Aenderung am Text ist ein
Verstoss.
Geprueft wird nur der aktuelle Push, nicht die Historie - Altlasten wuerden die
Pipeline dauerhaft rot faerben, ohne dass jemand sie beheben kann (FB-03).
Beide Sperren mit Positivkontrolle belegt: mein realer Verstoss 6652125 wird
rot (23 Zeilen), ein Wegwerf-Commit an docs/sources wird rot, ein sauberer
Commit gruen, ein ungueltiger Bereich rot.
⚠️ Beim Bau ist mir FB-02 prompt wieder unterlaufen: Der erste git-Helfer
verwandelte einen Fehlschlag in einen leeren String - ein ungueltiger Bereich
wurde damit zu 'keine Aenderungen' und bestand stillschweigend. Aufgefallen nur,
weil der Test gegen den Wurzel-Commit lief, der keinen Vorgaenger hat. Jetzt
wirft git() statt zu schweigen, und der Fall ist eine eigene Kontrolle.
GIT_DEPTH 0 fuer den validate-Job: Im flachen Klon faellt die Pruefung auf HEAD~1
zurueck und sagt das - genau messen ist besser als benannt lueckenhaft.
Von sorb benannt, im Repo nachgeprueft statt bestaetigt oder bestritten:
FB-08 Dokumente werden angelesen, nicht durchgelesen. roadmap.md wurde am
20.08. vollstaendig ausgegeben - 94 Zeilen - und die AAR-Pflicht in Zeile 94
nicht gesehen; sie steht dort seit dem 12.08. Danach wurde dieselbe Datei sogar
bearbeitet. Gleiches Muster: der .npmrc-Kommentar sagt den 404 woertlich voraus
und wurde erst nach dem Fehlschlag gelesen.
FB-09 Gelesene Anweisungen werden nicht befolgt. AGENTS.md Zeile 81 sagt, dass
ADRs nie editiert werden. Die Datei war gelesen.
FB-10 Gesperrte Dateien werden bearbeitet. Commit 6652125 hat die angenommene
ADR-0022 veraendert - committet und gepusht, zurueckgenommen erst mit 6731d2a,
und zwar nicht durch eine Pruefung, sondern durch Zufall beim Oeffnen der
Vorlage.
Die drei sind als Klasse 'Arbeitsweise' gefuehrt, nicht als Verfahrenslueck:
FB-09 ist die gefaehrlichste, weil das Wissen vorhanden war und nicht abgerufen
wurde. Der Vergleich, der es belegt: STATUS.md traegt dasselbe Verbot wie die
ADRs und wurde kein einziges Mal verletzt - weil gen_status --check
widerspricht.
Auf Wunsch von sorb: eine Eingangsliste fuer die Framework-Ernte, damit
wiederkehrende Fehlerklassen in der naechsten neckbeard-Iteration nicht erneut
passieren muessen. Je Befund kurze Beschreibung, Ursachen-Einschaetzung, Belege
und ein Vorschlag.
Name bewusst FRAMEWORK-BEFUNDE statt 'Framework Defects': 'Befund' ist im Projekt
der feststehende Begriff (gruppenpruefung meldet Befunde, quittungen quittiert
sie, AGENTS.md: 'Jeder Befund gehoert als Issue erfasst'). 'Defect' waere zudem
zu eng - das Uebergehen der AAR-Pflicht ist kein Defekt DES Frameworks, sondern
eine Abweichung, die eine Luecke darin sichtbar macht. Umbenennen ist ein git mv.
Sieben Erstbefunde, alle aus dieser Sitzung belegt:
FB-01 AAR-Pflicht setzt sich ohne Werkzeug nicht durch
FB-02 'meldet Erfolg, ist aber blind' - sieben Vorkommen, fuenf Werkzeuge
FB-03 Pruefungen erzeugen Befunde, die niemand beheben kann
FB-04 Issues behaupten Zustaende, die laengst ueberholt sind
FB-05 Dashboards und Pruefungen ohne Datenbeleg
FB-06 Dokumente werden ergaenzt, aber nicht revidiert
FB-07 Werkzeuge messen unbemerkt die falsche Instanz
Verlinkt in README (Struktur) und roadmap (Kadenz, bei der Retro) - eine Datei,
die niemand findet, waere selbst ein Befund der Klasse FB-06.
⚠️ validate.py prueft nur unterhalb von docs/. Diese Datei ist damit NICHT
schema-geprueft - bei einer Liste ueber fehlende Durchsetzung ist das eine
Ironie, die im Refinement einen Blick wert ist.
sorb hat zu Recht nachgehakt: Meine Aussage, nur der 19. und 20. seien
AAR-pflichtig gewesen, war unbelegt. Die Commits dieser beiden Tage zeigen
zwei weitere Vorfaelle.
1. Safari sendete nach der Freischaltung von v0.5.3 ungefiltert weiter -
LiveKits sender?.replaceTrack uebersprang den Tausch stumm, das rohe Mikrofon
blieb auf der Leitung. Datenschutzbezug: Wer den Filter einschaltete, durfte
annehmen, dass genau das nicht passiert.
2. Alle geplanten Pruefungen waren dauerhaft rot und meldeten damit nichts mehr.
Beide waren in ihren Issues ausfuehrlich dokumentiert - und beide ohne AAR.
Wichtiger als die Nacharbeit: #0054 nennt den Safari-Fall selbst das VIERTE
Vorkommen des Musters 'meldet Erfolg, ist aber blind'. Viermal erkannt, nie als
eigenstaendige Lehre abgelegt - und genau deshalb bin ich am 19./20. dreimal neu
hineingelaufen. Das Muster steht jetzt als Wiki-Seite mit sieben belegten
Vorkommen, Erkennungsmerkmalen und Gegenmitteln.
Der Validator hat die Seite prompt als verwaist gemeldet, bis sie im Index
verlinkt war. Wo ein Werkzeug widerspricht, halte ich den Prozess ein - das ist
exakt der Befund aus #0040.
Nicht als Vorfall gewertet und geprueft: die Boje-Aufraeumarbeit in #0074.
roadmap.md verlangt einen AAR nach jedem Deploy mit Uebergabe und nach
Incidents. Gestern und heute gab es vier solche Ereignisse - Produktionsausfall
durch rc.2, Verlust der DKIM-Eintraege, Game-Host-Deploy ueber zwei Uebergaben,
volle Wurzelpartition auf rohana mit gerissenem Spiegel. Geschrieben wurde
zunaechst keiner. Der letzte AAR im Repo stammte vom 16.08.
Dokumentiert habe ich stattdessen ausfuehrlich IN den Issues. Das ist nicht
wertlos, aber es ist nicht der Ort, an dem AGENTS.md die Lehren sucht - und die
Folge war messbar: dieselben Fallen wurden an einem Tag mehrfach neu entdeckt.
#0040 um eine Rueckmeldung ans Framework ergaenzt: Die Vorgaben wurden dort
eingehalten, wo ein Werkzeug widerspricht (validate, gen_status, Spiegel), und
uebergangen, wo nur Prosa sie verlangt. Mit Vorschlaegen fuers Refinement -
einschliesslich der ehrlichen Alternative, die Definition von Vorfall enger zu
fassen statt strenger zu kontrollieren.
Sperre beisst (wikijs-postgres und element-web-docs erreichen weder Internet noch
CFGMON), erlaubte Pfade funktionieren (DNS, Pod zu Postgres), breite Pods sind
unberuehrt (synapse erreicht rohana, CFGMON und 1.1.1.1).
Zwei eigene Fehlmessungen dokumentiert, weil sie wiederkommen: 10.43.0.1:443 ist
auch aus einem Pod mit breitem Ausgang nicht erreichbar, und /dev/tcp gibt es in
dash nicht - im Synapse-Container meldet es immer Fehlschlag.
Korrektur zur Zaehlung: zehn, nicht neun.
Nebenfund aus sorbs Frage zur URL-Vorschau: Synapses eigene Sperrliste enthaelt
169.254.0.0/16 nicht - heute durch die NetworkPolicy gedeckt, beim Umbau des
Egress-Modells aber nachzutragen.
Arbeitslasten durchgegangen: neun von zwanzig brauchen gar keinen Ausgang, sechs
genau ein Ziel (Storage Box, rohana), fuenf tatsaechlich breiten Zugang - Synapse
wegen aktiver URL-Vorschau, clamav wegen freshclam, coturn und SFU wegen Medien.
Wichtiger technischer Punkt: NetworkPolicies sind additiv. egress-block-metadata
waehlt ALLE Pods und erlaubt 0.0.0.0/0 - eine zusaetzliche enge Policy fuer
einen einzelnen Pod aendert daran nichts. Der Rueckbau muss das Modell drehen:
default-deny-egress plus ausdrueckliche Erlaubnis je Arbeitslast, so wie es
beim Ingress bereits ist.
Zuordnungen sind aus Konfiguration und Aufgabe abgeleitet, nicht gemessen - das
steht auch so im Issue.
#0083: Die drei im Issue genannten Dateien (prometheus.yml, alerts.yml,
alertmanager.yml) sind laengst Verzeichnis-Mounts - der vorgeschlagene Umbau ist
dort umgesetzt. Offen sind noch fuenf Mounts: loki-config.yaml, config.alloy und
drei .py-Skripte.
Bei den Skripten ist die Falle schlimmer, nicht harmloser: Python liest beim
Start, ein Neustart ist ohnehin noetig - und wird deshalb fuer ausreichend
gehalten. Der Prozess startet neu, der Mount zeigt weiter auf den alten Inode,
der Container meldet Erfolg und fuehrt alten Code aus. Ob die laufende
cve-exporter-Instanz die Aenderung vom 19.08. ausfuehrt, ist ungeprueft.
#0088: Beide Zahlen im Issue sind ueberholt - es sind 15 Ingress-Policies mit 21
Regeln, und eine Egress-Policy existiert inzwischen in drei Namespaces. Sie
sperrt aber genau ein Ziel: den Metadaten-Dienst. Alles andere darf weiterhin
ins Internet.
Der Kern des Befunds gilt damit weiter, nur die Zuspitzung 'unbeschraenkt'
nicht mehr. Beide Issues sagten Dinge, die so nicht mehr stimmen - genau die
Klasse, die heute schon mehrfach Zeit gekostet hat.
Zehn von zehn Targets up, 5 Server mit Klarnamen, Recording Rules liefern die
drei laufenden, Serienzahl von 3615 auf 1336 gefallen.
Die Ursache war nie die Firewall, wie ueber Wochen vermutet: Die Ports waren
gar nicht veroeffentlicht - nur im Docker-Netz erreichbar.
Der Exporter auf 9531 hatte nie funktioniert (Client-API mit Application-Key
ueber http), und das zugehoerige Dashboard war deshalb seit jeher leer, ohne
dass es jemandem auffiel. Beides ersetzt.
Zwei Messfallen festgehalten, weil sie wiederkommen: Der Entwicklungstunnel
fuehrt bei cfgmon.lab:9090 auf eine ANDERE Prometheus-Instanz, der echte
operating-Stack ist nur aus dem Cluster ueber 10.0.0.3 erreichbar. Und
--force-recreate ist fuer Prometheus ueberholt - das Verzeichnis wird gemountet.
Offen und unabhaengig: flux2.json nutzt gotk-Metriken, die es in dieser
Flux-Fassung nicht mehr gibt.
Aufraeumteil durch sorb erledigt (wiki-zugang geloescht, die vermutete tote App
gab es nicht). Letztes Abnahmekriterium umgesetzt: taeglicher CronJob zaehlt
Wiki-Nutzer ohne Gruppe und schreibt das Ergebnis nach Loki.
Beide Zweige geprueft - der Warn-Zweig laeuft nur, wenn etwas kaputt ist, also
genau dann, wenn er funktionieren muss.
Nebenfund mit Reichweite ueber dieses Issue hinaus: Die NetworkPolicy-Regeln fuer
einen neu erzeugten Pod sind beim Containerstart noch nicht programmiert. Im
selben Pod gemessen - sofort abgewiesen, nach 20 s erfolgreich. Betrifft
moeglicherweise auch wikijs-backup, synapse-backup und restore-drill.
Eigener TXT statt IONOS-CNAME, p=quarantine mit rua; sieben Service-Namen auf
Null-MX und -all; Apex unveraendert. Abnahme erfuellt - Authentik-Testmail kam
nachweislich an, nicht angenommen sondern geprueft.
Berichtsadresse ist gamemaster@axion1337.chat statt eines neuen Postfachs. Das
umgeht den Autorisierungs-Record, an dem eine zonenfremde Adresse still
gescheitert waere: ohne ihn senden die meisten Empfaenger gar nichts, und
ausbleibende Berichte sehen aus wie keine Probleme.
Zwischenfall festgehalten: Bei der Aenderung verschwanden alle drei
DKIM-Selektoren. Kein Ausfall - direkte Zustellung traegt ueber SPF -, aber
weitergeleitete Mail waere bei p=quarantine still im Spam gelandet. Wieder
aktiviert, s1/s2 loesen bis zum Schluessel auf.
Daraus zwei Werkzeug-Lehren, beide umgesetzt (notfallhandbuch 91d7106, b6ee7fd):
pruefe-dns.sh kannte DKIM nicht und haette den Verlust nicht gesehen; und es
misst im Lab den abgefangenen Resolver statt der Zone - eine Anfrage an eine
TEST-NET-Adresse lieferte ein Ergebnis. Beides jetzt abgedeckt, die Gegenprobe
laeuft vor den Messungen.
docker ps zeigt es eindeutig: Die Exporter veroeffentlichen ihre Ports gar nicht,
sind also nur im Docker-Netz erreichbar. Weder Firewall noch Bind-Adresse.
Zentrale Seite auf 10.0.0.4 umgestellt und der seit drei Monaten laufende
pterodactyl_exporter erstmals angebunden - das vorhandene Dashboard nutzt genau
dessen Metriken und war deshalb dauerhaft leer.
Host-Seite mit Reihenfolge und Stolpersteinen dokumentiert; Compose uebergeben.
#0077: ClamAV-Dashboard liegt in threadnet-operating (7ee42f9). Zwei Korrekturen
am Issue-Text, beide gemessen: das vorgeschlagene Label app existiert in dieser
Loki nicht (Cluster-Logs haengen an job=loki.source.kubernetes.k8s_logs mit
instance=<ns>/<pod>:<container>), und der Text kennt nur das Synapse-Modul -
die wichtigere Quelle ist der Client-Scanner, der auch verschluesselte Raeume
abdeckt. In sieben Tagen: 7 Treffer beim Client, 0 beim Modul.
Zusaetzlich ein Panel, das im Issue nicht stand: Fail-open-Faelle. Laesst der
Scanner mangels clamd Dateien durch, ist das sonst unsichtbar.
#0002: Der Host liegt inzwischen im vSwitch unter 10.0.0.4 - Ports 22/80/443
offen, 9100 und 8080 zu. Eine Umstellung der Targets auf die private IP genuegt
also NICHT; die Exporter binden nicht auf der privaten Schnittstelle. Der noetige
Handgriff liegt auf dem Game-Host. Mit Kontrollmessungen belegt, damit ein
stiller Fehlschlag nicht wie ein Befund aussieht.
Thore CimbalandClaude Opus 5