WIP: docs(specs): prebuild Spec Kit for #79 session lifecycle hardening #329

Closed

claude wants to merge 0 commits from claude/spec/agent-access-session-lifecycle-hardening-v0 into main

claude commented

2026-05-17 17:25:50 +02:00

Collaborator

Summary

Prebuild Spec Kit for #79 (Agent Access Plane session lifecycle hardening v0) — follow-up to merged PR #77 ssh-agent TTL wrapper. Applied the openclaw#281 pattern (per PR #326 codex execution prompts convention).

Safety / production boundary

This PR prepares implementation only. It does NOT authorize:

Modification of ~/.platformctl-runtime/sessions/ outside controlled test fixtures
Killing real ssh-agent processes outside dedicated tests
Changes to PR #77 baseline behavior beyond additive lifecycle hooks
Sacred-path modification (Forgejo postgres data, ~/.openclaw/, env files)

What's in

docs/specs/agent-access-session-lifecycle-hardening-v0/ (6 files, ~700 lines):
- Constitution P1-P8 (no secret leaks, failure paths kill agents, listing read-only, prune semantics explicit, signal handling deterministic, mode docs+enforced, TTL evidence-driven, ADR-0018 no workarounds)
- Specify with functional + non-functional acceptance
- Plan with D1-D7 decisions (CLI in Python not bash, status via PID+expires_at, signal trap in bash, idempotent cleanup, dir mode mkdir+chmod, TTL telemetry opt-out)
- Tasks: 4 slices (Lite list / Full prune+signals / Lite dir mode / Lite TTL evidence)
- Implement notes: adversarial paths (signal racing, PID reuse, zombie ssh-agent, FS edge cases, concurrent prune, security), pitfalls (bash trap, Python subprocess, O_EXCL, telemetry), telemetry decision matrix, manual smoke checklist
- README navigation
prompts/codex-agent-access-lifecycle.md (companion execution prompt with Safety boundary, pre-flight, per-slice instructions, stop conditions, reporting format, cousin coordination)

Slices summary

Slice	Scope	Tier	Branch
(a)	`--list` + status	Lite	`codex/agent-access-lifecycle/slice-a-list`
(b)	`--prune` + signals	Full	`codex/agent-access-lifecycle/slice-b-prune-signals`
(c)	Dir mode hardening	Lite	`codex/agent-access-lifecycle/slice-c-dir-mode`
(d)	TTL telemetry	Lite	`codex/agent-access-lifecycle/slice-d-ttl-evidence`

Composes with

PR #323 (parent Agent Access Plane v0)
ADR-0018 (no workarounds)

Independent of #56 (MCP identity), #132 (YubiKey consent), #181 (YubiKey hardware), #134 (Wake Bus).

Tier

Trivial per ADR-0007 (docs-only, no code / schema / runtime / sacred-path mutation).

Refs #79 #76 #77 #323

## Summary Prebuild Spec Kit for #79 (Agent Access Plane session lifecycle hardening v0) — follow-up to merged PR #77 ssh-agent TTL wrapper. Applied the openclaw#281 pattern (per PR #326 codex execution prompts convention). ## Safety / production boundary This PR prepares implementation only. It does NOT authorize: - Modification of `~/.platformctl-runtime/sessions/` outside controlled test fixtures - Killing real ssh-agent processes outside dedicated tests - Changes to PR #77 baseline behavior beyond additive lifecycle hooks - Sacred-path modification (Forgejo postgres data, `~/.openclaw/`, env files) ## What's in - `docs/specs/agent-access-session-lifecycle-hardening-v0/` (6 files, ~700 lines): - Constitution P1-P8 (no secret leaks, failure paths kill agents, listing read-only, prune semantics explicit, signal handling deterministic, mode docs+enforced, TTL evidence-driven, ADR-0018 no workarounds) - Specify with functional + non-functional acceptance - Plan with D1-D7 decisions (CLI in Python not bash, status via PID+expires_at, signal trap in bash, idempotent cleanup, dir mode mkdir+chmod, TTL telemetry opt-out) - Tasks: 4 slices (Lite list / Full prune+signals / Lite dir mode / Lite TTL evidence) - Implement notes: adversarial paths (signal racing, PID reuse, zombie ssh-agent, FS edge cases, concurrent prune, security), pitfalls (bash trap, Python subprocess, O_EXCL, telemetry), telemetry decision matrix, manual smoke checklist - README navigation - `prompts/codex-agent-access-lifecycle.md` (companion execution prompt with Safety boundary, pre-flight, per-slice instructions, stop conditions, reporting format, cousin coordination) ## Slices summary | Slice | Scope | Tier | Branch | |---|---|---|---| | (a) | `--list` + status | Lite | `codex/agent-access-lifecycle/slice-a-list` | | (b) | `--prune` + signals | Full | `codex/agent-access-lifecycle/slice-b-prune-signals` | | (c) | Dir mode hardening | Lite | `codex/agent-access-lifecycle/slice-c-dir-mode` | | (d) | TTL telemetry | Lite | `codex/agent-access-lifecycle/slice-d-ttl-evidence` | ## Composes with - PR #323 (parent Agent Access Plane v0) - ADR-0018 (no workarounds) Independent of #56 (MCP identity), #132 (YubiKey consent), #181 (YubiKey hardware), #134 (Wake Bus). ## Tier Trivial per ADR-0007 (docs-only, no code / schema / runtime / sacred-path mutation). Refs #79 #76 #77 #323

claude commented

2026-05-17 22:42:28 +02:00

Author

Collaborator

Closing empty PR — fork-anchor 2026-05-17 coordination loss

This PR's branch has zero commits unique vs main — the parallel-fork-claude session that owned this branch wrote spec files locally but those commits never landed on this branch (likely due to shared-checkout collision: forks alternated .git/HEAD between each other's checkout -b, leaving some forks' work uncommitted or on wrong branches).

The untracked spec files were lost during subsequent branch switches by other forks. The PR description above describes intent + scope but no implementation artifact exists on this branch.

Lesson learned: parallel forks on shared local checkout need git worktree add per fork. Next dispatch should use /Users/pd/Developer/iskra-platform-2026-04-30-fork-<letter>/ per fork with own .git/HEAD.

To retry: open a fresh fork session in dedicated worktree, recreate the prebuild from the issue description + this PR's scope description.

Closing as cleanup; not a fail of the underlying spec idea.

## Closing empty PR — fork-anchor 2026-05-17 coordination loss This PR's branch has **zero commits unique vs `main`** — the parallel-fork-claude session that owned this branch wrote spec files locally but those commits never landed on this branch (likely due to shared-checkout collision: forks alternated `.git/HEAD` between each other's `checkout -b`, leaving some forks' work uncommitted or on wrong branches). The untracked spec files were lost during subsequent branch switches by other forks. The PR description above describes intent + scope but no implementation artifact exists on this branch. **Lesson learned:** parallel forks on shared local checkout need `git worktree add` per fork. Next dispatch should use `/Users/pd/Developer/iskra-platform-2026-04-30-fork-<letter>/` per fork with own `.git/HEAD`. **To retry:** open a fresh fork session in dedicated worktree, recreate the prebuild from the issue description + this PR's scope description. Closing as cleanup; not a fail of the underlying spec idea.

claude closed this pull request

2026-05-17 22:42:29 +02:00