feat(autonomy): add fail-closed classifier gate #695

Merged

pdurlej merged 1 commit from codex/687-classifier-policy into main 2026-06-02 15:20:22 +02:00

Conversation 3 Commits 1 Files changed 3 +465

codex commented 2026-06-02 14:47:31 +02:00

Collaborator

Copy link

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

Summary

Adds the Tier-2 autonomy classifier interface for #687 without calling a model or mutating runtime.

This PR keeps the hard architectural guarantee intact: deterministic routing runs first, hard stops never reach the classifier, and missing/invalid/low-confidence classifier output fails closed to ask.

Canary Context Pack

Product story

The operator should not be interrupted for every medium-risk agent action, but irreversible actions must still remain deterministic operator gates. This slice gives Codex/agents a policy-as-text classifier packet for medium actions without weakening hard stops.

What changed

• Added a fail-closed classifier result/response layer in platformctl.autonomy.
• Added deterministic policy-as-text prompt compilation with policy digesting and secret-like redaction.
• Added platformctl autonomy ask for local classifier packet rendering and optional JSON classifier-response validation.
• Added router/classifier/CLI regression tests.

Why it changed

#687 is implementing the tiered autonomy gate in slices. PR #689 added the cascade router and PR #694 added apply sandbox receipt mode. This PR adds the Tier-2 classifier interface while still depending on an external cheap model later instead of building one now.

Files touched

• control-plane/platformctl/autonomy.py
• control-plane/platformctl/cli.py
• control-plane/platformctl/tests/test_autonomy_router.py

Relevant context

• state/strategy/autonomy-tiered-execution-design-2026-06-02.md
• Issue #687
• PR #689 cascade router
• PR #694 apply sandbox receipt mode

Runtime evidence

No runtime action, no model call, no Forgejo mutation outside this PR creation. Local validation:

• PYTHONPATH=control-plane python3 -m pytest control-plane/platformctl/tests/test_autonomy_router.py → 15 passed
• PYTHONPATH=control-plane python3 -m pytest control-plane/platformctl/tests/test_apply_phase3.py control-plane/platformctl/tests/test_apply_env_file.py control-plane/platformctl/tests/test_autonomy_router.py → 134 passed
• PYTHONPATH=control-plane python3 -m platformctl.cli validate all --json → exitCode 0

Known constraints

The classifier interface does not call Ollama/OpenAI/Claude. It accepts an optional normalized response artifact and otherwise returns ask. Decision receipts into ADR-0025 memory plane remain a later #687 slice.

Explicit out-of-scope

• No live model invocation.
• No runtime apply.
• No live Postgres migration.
• No issue/comment writer.
• No decision receipt persistence yet.

Requested decision

Approve if the fail-closed classifier packet shape and CLI are acceptable as the Tier-2 interface.

Merge blockers

• Any path where hard-stop actions reach the classifier.
• Any path where missing/invalid/low-confidence classifier output returns allow.
• Any raw secret-like context leaking into the generated classifier prompt.

Spec sources read

• state/strategy/autonomy-tiered-execution-design-2026-06-02.md — #687 design and classifier contract.
• control-plane/platformctl/autonomy.py — existing cascade router implementation.
• control-plane/platformctl/tests/test_autonomy_router.py — existing router regression coverage.
• control-plane/platformctl/cli.py — CLI registration conventions.
• control-plane/platformctl/apply.py — previous sandbox receipt event shape and redaction precedent.

Refs #687

Canary status: missing — fire canary 3+3 manually before merge ## Summary Adds the Tier-2 autonomy classifier interface for #687 without calling a model or mutating runtime. This PR keeps the hard architectural guarantee intact: deterministic routing runs first, hard stops never reach the classifier, and missing/invalid/low-confidence classifier output fails closed to `ask`. ## Canary Context Pack ### Product story The operator should not be interrupted for every medium-risk agent action, but irreversible actions must still remain deterministic operator gates. This slice gives Codex/agents a policy-as-text classifier packet for medium actions without weakening hard stops. ### What changed - Added a fail-closed classifier result/response layer in `platformctl.autonomy`. - Added deterministic policy-as-text prompt compilation with policy digesting and secret-like redaction. - Added `platformctl autonomy ask` for local classifier packet rendering and optional JSON classifier-response validation. - Added router/classifier/CLI regression tests. ### Why it changed #687 is implementing the tiered autonomy gate in slices. PR #689 added the cascade router and PR #694 added apply sandbox receipt mode. This PR adds the Tier-2 classifier interface while still depending on an external cheap model later instead of building one now. ### Files touched - `control-plane/platformctl/autonomy.py` - `control-plane/platformctl/cli.py` - `control-plane/platformctl/tests/test_autonomy_router.py` ### Relevant context - `state/strategy/autonomy-tiered-execution-design-2026-06-02.md` - Issue #687 - PR #689 cascade router - PR #694 apply sandbox receipt mode ### Runtime evidence No runtime action, no model call, no Forgejo mutation outside this PR creation. Local validation: - `PYTHONPATH=control-plane python3 -m pytest control-plane/platformctl/tests/test_autonomy_router.py` → 15 passed - `PYTHONPATH=control-plane python3 -m pytest control-plane/platformctl/tests/test_apply_phase3.py control-plane/platformctl/tests/test_apply_env_file.py control-plane/platformctl/tests/test_autonomy_router.py` → 134 passed - `PYTHONPATH=control-plane python3 -m platformctl.cli validate all --json` → exitCode 0 ### Known constraints The classifier interface does not call Ollama/OpenAI/Claude. It accepts an optional normalized response artifact and otherwise returns `ask`. Decision receipts into ADR-0025 memory plane remain a later #687 slice. ### Explicit out-of-scope - No live model invocation. - No runtime apply. - No live Postgres migration. - No issue/comment writer. - No decision receipt persistence yet. ### Requested decision Approve if the fail-closed classifier packet shape and CLI are acceptable as the Tier-2 interface. ### Merge blockers - Any path where hard-stop actions reach the classifier. - Any path where missing/invalid/low-confidence classifier output returns `allow`. - Any raw secret-like context leaking into the generated classifier prompt. ## Spec sources read - `state/strategy/autonomy-tiered-execution-design-2026-06-02.md` — #687 design and classifier contract. - `control-plane/platformctl/autonomy.py` — existing cascade router implementation. - `control-plane/platformctl/tests/test_autonomy_router.py` — existing router regression coverage. - `control-plane/platformctl/cli.py` — CLI registration conventions. - `control-plane/platformctl/apply.py` — previous sandbox receipt event shape and redaction precedent. Refs #687

codex added 1 commit 2026-06-02 14:47:31 +02:00

feat(autonomy): add fail-closed classifier gate 4d8594af8c

codex added the

class/security-sensitive

label 2026-06-02 14:47:55 +02:00

codex commented 2026-06-02 14:55:10 +02:00

Author

Collaborator

Copy link

Patchwarden PR sanity

• Status: advisory_findings
• PR: 695
• Commit: 2521359bc913e74787fce7c76279c591cb0ae560
• Security-sensitive label: present
• Authority: advisory model review plus deterministic blockers only
• 3+3 canary: still alive; this does not replace it

Deterministic findings

No deterministic findings.

Model reviewers

global-glm / glm-5.1:cloud

• Status: ok
• Verdict: OK
• Findings: none

global-deepseek / deepseek-v4-pro:cloud

• Status: ok

• Verdict: OK

• medium Incomplete secret redaction may leak sensitive data in classifier prompt

  • Evidence: control-plane/platformctl/autonomy.py: _SECRET_REPLACEMENTS regex patterns cover only Bearer tokens, 40-char hex strings, olostep_ keys, and key=value assignments. Other common secret formats (e.g., AWS keys, JWT, shorter hex tokens) are no
  • Next: Expand redaction patterns to cover additional secret formats or add a prominent warning that the generated prompt may contain sensitive data and must not be logged or stored without further sanitization.

redteam / kimi-k2.6:cloud

• Status: ok

• Verdict: NOT_OK

• blocker NaN confidence bypasses fail-closed check for direct AutonomyClassifierResponse objects

  • Evidence: control-plane/platformctl/autonomy.py: classify_action accepts a pre-constructed AutonomyClassifierResponse and uses it directly without validation (bypassing from_mapping). The confidence gate response.confidence < min_confidence evaluat
  • Next: Add post_init validation to AutonomyClassifierResponse to enforce decision in {allow,retry,ask} and confidence is finite and within [0,1], or explicitly validate response.confidence with math.isfinite before the confidence gate in classify_action.

Policy notes

• GLM 5.1 + DeepSeek V4 Pro are the operator-required model mix for this bot.
• Optional red-team model is enabled only when PLATFORMCTL_PR_SANITY_REDTEAM_MODEL is configured.
• Auto-merge is not enabled here.

<!-- patchwarden-pr-sanity:pdurlej/platform:PR-695 --> # Patchwarden PR sanity - Status: `advisory_findings` - PR: `695` - Commit: `2521359bc913e74787fce7c76279c591cb0ae560` - Security-sensitive label: `present` - Authority: advisory model review plus deterministic blockers only - 3+3 canary: still alive; this does not replace it ## Deterministic findings No deterministic findings. ## Model reviewers ### `global-glm` / `glm-5.1:cloud` - Status: `ok` - Verdict: `OK` - Findings: none ### `global-deepseek` / `deepseek-v4-pro:cloud` - Status: `ok` - Verdict: `OK` - **`medium`** Incomplete secret redaction may leak sensitive data in classifier prompt - Evidence: `control-plane/platformctl/autonomy.py: _SECRET_REPLACEMENTS regex patterns cover only Bearer tokens, 40-char hex strings, olostep_ keys, and key=value assignments. Other common secret formats (e.g., AWS keys, JWT, shorter hex tokens) are no` - Next: Expand redaction patterns to cover additional secret formats or add a prominent warning that the generated prompt may contain sensitive data and must not be logged or stored without further sanitization. ### `redteam` / `kimi-k2.6:cloud` - Status: `ok` - Verdict: `NOT_OK` - **`blocker`** NaN confidence bypasses fail-closed check for direct AutonomyClassifierResponse objects - Evidence: `control-plane/platformctl/autonomy.py: classify_action accepts a pre-constructed AutonomyClassifierResponse and uses it directly without validation (bypassing from_mapping). The confidence gate `response.confidence < min_confidence` evaluat` - Next: Add __post_init__ validation to AutonomyClassifierResponse to enforce decision in {allow,retry,ask} and confidence is finite and within [0,1], or explicitly validate response.confidence with math.isfinite before the confidence gate in classify_action. ## Policy notes - GLM 5.1 + DeepSeek V4 Pro are the operator-required model mix for this bot. - Optional red-team model is enabled only when `PLATFORMCTL_PR_SANITY_REDTEAM_MODEL` is configured. - Auto-merge is not enabled here.

codex force-pushed codex/687-classifier-policy from 4d8594af8c to 11f2ddd12e 2026-06-02 14:59:46 +02:00 Compare

codex force-pushed codex/687-classifier-policy from 11f2ddd12e to 2521359bc9 2026-06-02 15:09:17 +02:00 Compare

pdurlej approved these changes 2026-06-02 15:20:21 +02:00

pdurlej left a comment

Copy link

Approved after Patchwarden security-sensitive sanity passed with 12/12 green checks. Automated Codex operator-approved lane; no runtime/destructive action in this PR.

Approved after Patchwarden security-sensitive sanity passed with 12/12 green checks. Automated Codex operator-approved lane; no runtime/destructive action in this PR.

pdurlej approved these changes 2026-06-02 15:20:21 +02:00

pdurlej left a comment

Copy link

Approved after Patchwarden security-sensitive sanity passed with 12/12 green checks. Automated Codex operator-approved lane; no runtime/destructive action in this PR.

Approved after Patchwarden security-sensitive sanity passed with 12/12 green checks. Automated Codex operator-approved lane; no runtime/destructive action in this PR.

pdurlej merged commit 3af827d0ec into main 2026-06-02 15:20:22 +02:00

pdurlej referenced this pull request from a commit 2026-06-02 15:20:22 +02:00

Merge pull request #695 from codex/687-classifier-policy

codex referenced this pull request 2026-06-02 15:24:53 +02:00

feat(autonomy): write decision receipts #696

Sign in to join this conversation.

Reviewers

No reviewers

pdurlej

Labels

Clear labels   

W6d-automerge-calibration

  

agent/claude-code

Vistula actor: Claude Code.

  

agent/codex

Vistula actor: Codex.

  

agent/hermes

Vistula actor: Hermes Upstream.

  

agent/iskra

Vistula actor: Iskra.

  

agent/ollama

Vistula actor: Ollama or compatible model.

  

agent/patchwarden

Vistula actor: Patchwarden.

  

automerge-candidate

Candidate for narrow W6d autonomous merge readiness pilot

  

class/security-sensitive

Security-sensitive class of service: smallest coherent PRs and full canary

  

cutover-gate

blocks production declare / final RS2000 replacement

  

dependency/blocked

Vistula dependency state: blocked.

  

dependency/blocks-others

Vistula dependency state: blocks other work.

  

dependency/cross-repo

Vistula dependency scope: cross-repository.

  

dependency/needs-confirmation

Vistula dependency state: needs confirmation.

  

domain:agents

Agent identity, delegation, subagents, routing, or workflows.

  

domain:ci

CI, checks, Actions, runners, or release automation.

  

domain:docs

Documentation, packets, practices, or source-of-truth text.

  

domain:forgejo

Forgejo issues, PRs, labels, repo settings, or identity.

  

domain:infra

Infrastructure, storage, backup, DNS, network, or host substrate.

  

domain:memory

Memory, recall, write gates, salience, or Honcho-adjacent behavior.

  

domain:runtime

Runtime services, deploys, daemons, timers, or live behavior.

  

domain:signal

Signal UX, delivery, context, or outbound behavior.

  

domain:ux

Product UX, operator experience, or conversation ergonomics.

  

flow/architecture

Vistula lifecycle: architecture phase in progress.

  

flow/blocked

Vistula lifecycle: blocked on dependency or decision.

  

flow/deployed

Vistula lifecycle: deployed to runtime where applicable.

  

flow/done

Vistula lifecycle: work merged or otherwise done.

  

flow/implementation

Vistula lifecycle: implementation phase in progress.

  

flow/intake

Vistula lifecycle: incoming signal not refined yet.

  

flow/maintained

Vistula lifecycle: maintained/steady state.

  

flow/observed

Vistula lifecycle: observed after delivery.

  

flow/ready

Vistula lifecycle: ready for architecture or implementation.

  

flow/refining

Vistula lifecycle: Hermes/spec refinement in progress.

  

flow/retired

Vistula lifecycle: retired or intentionally closed.

  

flow/review

Vistula lifecycle: review phase in progress.

  

iterating

Orchestrator/Codex actively amending in response to review — operator can ignore until resolved

  

judge/codex-candidate

Candidate for Codex implementation.

  

judge/hermes-candidate

Candidate for Hermes discovery/research.

  

judge/low-confidence

Judgment confidence is low.

  

judge/needs-refinement

Needs clearer evidence or scope.

  

judge/operator-needed

Needs Piotr decision or input.

  

judge/p0

Urgent, operator attention or unblock now.

  

judge/p1

High value, schedule soon.

  

judge/p2

Useful, normal queue.

  

judge/p3

Low priority, keep but do not chase.

  

judge/park

Parked from current attention.

  

judge/patchwarden-candidate

Needs Patchwarden merge-safety attention.

  

judge/stale-priority

Existing priority judgment appears stale.

  

kind/adr

Vistula work kind: architecture decision record.

  

kind/bug

Vistula work kind: bug fix.

  

kind/chore

Vistula work kind: chore.

  

kind/feature

Vistula work kind: feature.

  

kind/infra

Vistula work kind: infrastructure.

  

kind/ops

Vistula work kind: operations.

  

kind/refactor

Vistula work kind: refactor.

  

kind/research

Vistula work kind: research/discovery.

  

large-impact

Per charter §3 review-by-impact: large impact (≥300 LOC OR new module OR identity/secret change OR architectural pivot) — Oracle pre-review recommended

  

merge/auto

Vistula merge mode: automated merge.

  

merge/manual

Vistula merge mode: manual merge.

  

merge/manual-dependency-conflict

Manual merge reason: dependency conflict.

  

merge/manual-failing-tests

Manual merge reason: failing tests.

  

merge/manual-merge-conflict

Manual merge reason: merge conflict.

  

merge/manual-missing-review

Manual merge reason: missing review.

  

merge/manual-operator-preference

Manual merge reason: operator preference.

  

merge/manual-red-zone

Manual merge reason: red-zone risk.

  

merge/manual-security-sensitive

Manual merge reason: security-sensitive work.

  

merge/manual-unclear-scope

Manual merge reason: unclear scope.

  

merge/manual-unknown

Manual merge reason: unknown.

  

meta

meta-decomposition issue; first agent decomposes into atomic children

  

mode:operator-only

Apply step requires explicit operator approval.

  

mode:patchwarden-iskra-approved

Work eligible for Patchwarden/Iskra approval inside repo policy.

  

mode:safe-auto

Low-risk work eligible for lightweight autonomous handling.

  

needs-operator-decision

PR is blocked on operator yes/no decision — fresh-Claude/Codex won't make this call alone

  

needs-triage

decomposing agent has open questions / unknowns

  

not-ready

reviewed but blocked on a precondition

  

observed/erroring

Vistula observation: erroring.

  

observed/needs-followup

Vistula observation: needs follow-up.

  

observed/pending

Vistula observation: pending.

  

observed/retire-candidate

Vistula observation: retire candidate.

  

observed/unused

Vistula observation: unused.

  

observed/used

Vistula observation: used.

  

operator-emotional

operator-emotional importance (Iskra memory etc.)

  

owner-attention

owner must decide; blocks default path

  

phase/02

Phase 02 cataloging

  

phase/03

Phase 03 control plane (platformctl)

  

priority:p0

Immediate safety, uptime, data, or operator-blocking issue.

  

priority:p1

Important active work; should be pulled soon.

  

priority:p2

Useful normal-priority work.

  

priority:p3

Opportunistic cleanup, research, or backlog work.

  

proposed

auto-generated by meta-decomposition; awaiting human review

  

ready-for-agent

reviewed and pickable; agents may claim per cookbook

  

ready-for-operator

PR is ready for operator review/merge — orchestrator finished its part

  

recovery

disaster recovery / restore semantics relevant

  

review:claude-reviewed

Reviewed by Claude-family agent.

  

review:codex-reviewed

Reviewed by Codex-family agent.

  

review:dziadek-reviewed

Reviewed by Dziadek pinch-point reviewer.

  

review:needs-human

Needs human review before merge or apply.

  

risk/exposure

incorrectly public/private/tailnet/auth/routed

  

risk/process

workflow/review/orchestration may produce bad outcomes

  

risk/product

work may have lost owner/user value

  

risk/runtime

platform may not run, recover, validate, behave reproducibly

  

safety:external-write

May write to external systems, users, rooms, repos, or services.

  

safety:no-prod-mutation

Must not mutate production or live external state.

  

safety:prod-impact

May affect production/runtime behavior.

  

safety:secret-touch

Touches secrets, credentials, auth, or secret-adjacent config.

  

size/large

Vistula size: large.

  

size/medium

Vistula size: medium.

  

size/small

Vistula size: small.

  

size/tiny

Vistula size: tiny.

  

size/unknown

Vistula size: not classified yet.

  

source/adr

Vistula source: ADR.

  

source/agent-generated

Vistula source: agent generated.

  

source/manual

Vistula source: manual entry.

  

source/operator-chat

Vistula source: operator chat.

  

source/voice-note

Vistula source: voice note.

  

status:blocked

Blocked by missing dependency, evidence, access, or operator decision.

  

status:codex-ready

Ready for a Codex implementation pass.

  

status:merged:pending-evidence

Merged but waiting for live or artifact evidence before closure.

  

status:needs-evidence

Needs proof, logs, tests, or operator-visible evidence.

  

status:operator-needed

Requires explicit operator input before safe progress.

  

status:parked

Intentionally deferred; not active campaign work.

  

tier/full

Full review tier per ADR-0007

  

tier/lite

Lite review tier per ADR-0007

  

tier/stacked

Stacked PR (base != main) escape hatch per ADR-0017. Requires consolidator-PR plan in PR body.

  

tier:0-platform-substrate

Data, secrets, backup, storage, or critical platform substrate.

  

tier:1-iskra-value-layer

Iskra/OpenClaw behavior, memory, runtime, or value-layer work.

  

tier:2-tools-products-modules

Tools, products, modules, experiments, and lower-blast-radius work.

  

type:bug

Incorrect behavior or regression.

  

type:chore

Maintenance, cleanup, metadata, or operational task.

  

type:docs

Documentation-only work.

  

type:feat

New capability or product behavior.

  

type:policy

Governance, protocol, boundary, or decision-contract change.

  

type:research

Investigation, spike, or evidence-gathering task.

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

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

2 participants

Notifications

Subscribe

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!695

No description provided.