Commit Graph
2 Commits
Author SHA1 Message Date
jakman85andClaude Opus 5 6253d546f8 docs(pm-agent): the registry cannot be pulled from yet — name both prerequisites where the operator reads them
gate / gate (pull_request) Successful in 1m50s
gate / conformance (pull_request) Has been skipped
pm-agent reconciled exactly as designed after #57: Flux applied main@sha1:95597bd to
platform-config, pm-agent and apps-wave-0 (all Ready), the ClusterPolicy
pm-agent-job-shape is live with Enforce + failurePolicy Fail and all six rules, SOPS
decryption works through the new overlay patch (checked by key names and value lengths,
never values), and pm-agent-worker really is bound to nothing.

Measuring it turned up two prerequisites for the NEXT operator step that nothing in this
repository declares, and that #57's handoff could not have known — it listed the image
pull as unverified, and that is exactly where it bit:

 - the nodes cannot verify the registry's TLS certificate (containerd: 'x509: certificate
   is not valid for any names'), while the same name serves a healthy Let's Encrypt
   wildcard from the LAN. A correct digest fails identically;
 - an unauthenticated client gets nothing from this Gitea's registry (its anonymous token
   carries no access scope) and no imagePullSecret exists anywhere in the tree.

Not specific to pm-agent: conformance-runner shares the registry path and has never been
pulled either. The dispatcher is the first workload in the cluster ever to reference it.

Documentation only — no manifest, no policy, no code. Both fixes need an operator action
and one of them is a posture decision, so the docs present both rather than pick.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
2026-10-05 23:22:52 -05:00
jakman85andClaude Opus 5 c5b5b8672a feat(pm-agent): an in-cluster AI Program Manager that turns Gitea issues into pull requests
gate / gate (pull_request) Failing after 1m51s
gate / conformance (pull_request) Has been skipped
A service that reads this repository's Gitea issues, decides what to do about each
thread as the Program Manager persona, and runs one worker agent per assignment that
authors a change, runs the merge gate and opens a pull request. It reacts to issue
creation and to comments; the assignment and all correspondence travel through the
issue thread and the PR. It authenticates with an Anthropic subscription token. It
never pushes to main and never merges.

Three choices carry the design, with alternatives recorded in docs/plans/:
 - it POLLS rather than receiving webhooks (Gitea's webhook SSRF guard, lost
   deliveries, and no inbound reach needed at all);
 - it keeps NO state: the trigger is "the newest thing in this thread was not
   written by me", so the issue thread is the state — no PVC, no backup obligation;
 - the model holds NO tools: its reply is one JSON object this code validates, so
   nothing in the path executes text.

The boundary is platform/kyverno/admission/pm-agent-job-shape.yaml (Enforce,
failurePolicy: Fail), not the RBAC: Kubernetes places no `use` check on
serviceAccountName, so "may create Jobs" is "may run arbitrary code as the
namespace's most privileged identity". Six rules admit only the committed shape, and
the worker identity is bound to nothing with no token mounted, so there is nothing to
escalate to. 28/28 offline Kyverno assertions.

Nothing has run: the image has never been built or pushed, and both the image digest
and the two credentials are deliberate placeholders. See the handoff entry's
"What I did NOT verify".

Refs CLAUDE.md §1/§3/§6, R-SEC-02, R-GITOPS-02.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
2026-10-05 22:28:10 -05:00