fix(plan): allow host-agent repo digest observation #298

Merged
pdurlej merged 1 commit from codex/plan/host-agent-repodigests-allowlist into main 2026-05-16 12:53:37 +02:00
Collaborator

Canary status: missing - fire canary 3+3 manually before merge

Canary Context Pack

Product story

The canonical Meerkat frontend smoke is blocked by a false image drift: platformctl plan now knows how to compare registry RepoDigests, but the RS2000 host-agent forced-command wrapper still rejects the read-only docker image inspect command needed to collect them.

What changed

  • Allow platform-host-agent to run only docker image inspect sha256:<64> --format '{{json .RepoDigests}}.
  • Keep tag-based image inspection and arbitrary formats denied.
  • Add a plan diagnostic when RepoDigest observation fails, so this failure mode is visible in artifacts instead of silently degrading to false drift.

Why it changed

Run API 1202 / UI #946 picked up immediately on runner 5, but artifact still showed Meerkat image drift with candidates limited to the tag and local image id. Direct runner-path test showed platform-host-agent: denied for the new digest lookup command.

Files touched

  • ops/rs2000/platform-host-agent-wrapper
  • tests/test_platform_host_agent_wrapper.py
  • control-plane/platformctl/plan.py
  • control-plane/platformctl/tests/test_plan_phase3.py

Relevant context

  • PR #297 added RepoDigest observation to platformctl plan.
  • PR #296 imported Meerkat canonical compose.
  • Issue #142 RS2000 cutover lane.

Runtime evidence

  • Meerkat retry after #297: run API 1202, UI #946, immediate runner pickup, no watchdog restart.
  • Artifact platformctl-auto-apply-1202 showed status=drift, observed=[tag, sha256:image-id], no RepoDigests.
  • Runner-path command returned platform-host-agent: denied for docker image inspect sha256:... --format '{{json .RepoDigests}}'.

Known constraints

This is a host forced-command allowlist. The wrapper remains narrow: no shell metacharacters, no tag lookup, no arbitrary image inspect format, no compose widening.

Explicit out-of-scope

  • No F3 smoke in this PR.
  • No production wrapper installation in the PR itself.
  • No direct PAT/Infisical changes.

Requested decision

Merge this Full/security-sensitive fix, then install the audited wrapper on RS2000 and retry np-meerkat-frontend before F3.

Merge blockers

  • The wrapper must not allow arbitrary image inspection or tag-based probing.
  • Tests must pass.

Spec sources read

  • ops/rs2000/platform-host-agent-wrapper - live forced-command policy mirrored in repo.
  • tests/test_platform_host_agent_wrapper.py - wrapper contract tests.
  • control-plane/platformctl/plan.py - RepoDigest observation path.
  • control-plane/platformctl/tests/test_plan_phase3.py - plan observation tests.

Verification

  • bash -n ops/rs2000/platform-host-agent-wrapper
  • git diff --check
  • uv run --project control-plane pytest tests/test_platform_host_agent_wrapper.py control-plane/platformctl/tests/test_plan_phase3.py -> 23 passed

Refs: #142, #296, #297

Canary status: missing - fire canary 3+3 manually before merge ## Canary Context Pack ### Product story The canonical Meerkat frontend smoke is blocked by a false image drift: `platformctl plan` now knows how to compare registry RepoDigests, but the RS2000 host-agent forced-command wrapper still rejects the read-only `docker image inspect` command needed to collect them. ### What changed - Allow `platform-host-agent` to run only `docker image inspect sha256:<64> --format '{{json .RepoDigests}}`. - Keep tag-based image inspection and arbitrary formats denied. - Add a plan diagnostic when RepoDigest observation fails, so this failure mode is visible in artifacts instead of silently degrading to false drift. ### Why it changed Run API `1202` / UI `#946` picked up immediately on runner `5`, but artifact still showed Meerkat image drift with candidates limited to the tag and local image id. Direct runner-path test showed `platform-host-agent: denied` for the new digest lookup command. ### Files touched - `ops/rs2000/platform-host-agent-wrapper` - `tests/test_platform_host_agent_wrapper.py` - `control-plane/platformctl/plan.py` - `control-plane/platformctl/tests/test_plan_phase3.py` ### Relevant context - PR #297 added RepoDigest observation to `platformctl plan`. - PR #296 imported Meerkat canonical compose. - Issue #142 RS2000 cutover lane. ### Runtime evidence - Meerkat retry after #297: run API `1202`, UI `#946`, immediate runner pickup, no watchdog restart. - Artifact `platformctl-auto-apply-1202` showed `status=drift`, `observed=[tag, sha256:image-id]`, no RepoDigests. - Runner-path command returned `platform-host-agent: denied` for `docker image inspect sha256:... --format '{{json .RepoDigests}}'`. ### Known constraints This is a host forced-command allowlist. The wrapper remains narrow: no shell metacharacters, no tag lookup, no arbitrary image inspect format, no compose widening. ### Explicit out-of-scope - No F3 smoke in this PR. - No production wrapper installation in the PR itself. - No direct PAT/Infisical changes. ### Requested decision Merge this Full/security-sensitive fix, then install the audited wrapper on RS2000 and retry `np-meerkat-frontend` before F3. ### Merge blockers - The wrapper must not allow arbitrary image inspection or tag-based probing. - Tests must pass. ## Spec sources read - `ops/rs2000/platform-host-agent-wrapper` - live forced-command policy mirrored in repo. - `tests/test_platform_host_agent_wrapper.py` - wrapper contract tests. - `control-plane/platformctl/plan.py` - RepoDigest observation path. - `control-plane/platformctl/tests/test_plan_phase3.py` - plan observation tests. ## Verification - `bash -n ops/rs2000/platform-host-agent-wrapper` - `git diff --check` - `uv run --project control-plane pytest tests/test_platform_host_agent_wrapper.py control-plane/platformctl/tests/test_plan_phase3.py` -> 23 passed Refs: #142, #296, #297
fix(plan): allow host-agent repo digest observation
All checks were successful
base-is-main / guard (pull_request) Successful in 2s
canary-required / collect-diff (pull_request) Successful in 4s
patchwarden-pr-sanity / collect-diff (pull_request) Successful in 3s
pyfallow / Pyfallow gate (control-plane) (pull_request) Successful in 18s
python-ci / Python 3.12 (pull_request) Successful in 37s
python-ci / Python 3.13 (pull_request) Successful in 38s
platformctl plan / auto-apply scope (pull_request) Successful in 21s
python-ci / Python 3.11 (pull_request) Successful in 37s
patchwarden-pr-sanity / sanity (pull_request) Successful in 19s
canary-required / canary (pull_request) Successful in 13s
ea0a9d55bb
Sign in to join this conversation.
No reviewers
No labels
W6d-automerge-calibration
agent/claude-code
agent/codex
agent/hermes
agent/iskra
agent/ollama
agent/patchwarden
automerge-candidate
class/security-sensitive
cutover-gate
dependency/blocked
dependency/blocks-others
dependency/cross-repo
dependency/needs-confirmation
domain:agents
domain:ci
domain:docs
domain:forgejo
domain:infra
domain:memory
domain:runtime
domain:signal
domain:ux
flow/architecture
flow/blocked
flow/deployed
flow/done
flow/implementation
flow/intake
flow/maintained
flow/observed
flow/ready
flow/refining
flow/retired
flow/review
iterating
judge/codex-candidate
judge/hermes-candidate
judge/low-confidence
judge/needs-refinement
judge/operator-needed
judge/p0
judge/p1
judge/p2
judge/p3
judge/park
judge/patchwarden-candidate
judge/stale-priority
kind/adr
kind/bug
kind/chore
kind/feature
kind/infra
kind/ops
kind/refactor
kind/research
large-impact
merge/auto
merge/manual
merge/manual-dependency-conflict
merge/manual-failing-tests
merge/manual-merge-conflict
merge/manual-missing-review
merge/manual-operator-preference
merge/manual-red-zone
merge/manual-security-sensitive
merge/manual-unclear-scope
merge/manual-unknown
meta
mode:operator-only
mode:patchwarden-iskra-approved
mode:safe-auto
needs-operator-decision
needs-triage
not-ready
observed/erroring
observed/needs-followup
observed/pending
observed/retire-candidate
observed/unused
observed/used
operator-emotional
owner-attention
phase/02
phase/03
priority:p0
priority:p1
priority:p2
priority:p3
proposed
ready-for-agent
ready-for-operator
recovery
review:claude-reviewed
review:codex-reviewed
review:dziadek-reviewed
review:needs-human
risk/exposure
risk/process
risk/product
risk/runtime
safety:external-write
safety:no-prod-mutation
safety:prod-impact
safety:secret-touch
size/large
size/medium
size/small
size/tiny
size/unknown
source/adr
source/agent-generated
source/manual
source/operator-chat
source/voice-note
status:blocked
status:codex-ready
status:merged:pending-evidence
status:needs-evidence
status:operator-needed
status:parked
tier/full
tier/lite
tier/stacked
tier:0-platform-substrate
tier:1-iskra-value-layer
tier:2-tools-products-modules
type:bug
type:chore
type:docs
type:feat
type:policy
type:research
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
pdurlej/platform!298
No description provided.