pdurlej/platform

Fork 0

test(platformctl): add provenance adversarial coverage #641

Merged

pdurlej merged 1 commit from codex/m06-provenance-adversarial-tests into main

2026-05-30 15:34:53 +02:00

codex commented

2026-05-30 15:24:22 +02:00

Collaborator

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

Canary Context Pack

Product story

The apply pipeline should fail closed under hostile provenance inputs, not only under happy-path malformed values.

What changed

Added adversarial coverage for corrupted plan JSON, type-confused plan payloads, forged clean env state, type-confused source SHA values, and fake manifest git roots.

Why it changed

Issue #194 closes the redaction/provenance hardening segment after #199, #200, #206, #192, and #193. These tests document the threat model before moving to plan-state/runtime-safety tasks.

Files touched

control-plane/platformctl/apply.py
control-plane/platformctl/tests/test_apply_phase3.py
control-plane/platformctl/tests/test_plan_phase3.py

Relevant context

Issue #194: PROVENANCE-ADVERSARIAL-TESTS-01
#199/#200/#206 redaction chain merged
#192/#193 provenance primitives merged

Runtime evidence

No runtime mutation. Local validation only.

Known constraints

This is intentionally test-heavy. The only code behavior change is fail-closed handling for invalid/non-object plan JSON before approval lookup.

Explicit out-of-scope

No plan-state changes (#196/#197 next)
No runtime apply or service restart
No destructive operations

Requested decision

Approve merge if tests and Patchwarden checks are green.

Merge blockers

Corrupted/type-confused plan JSON reaches approval/runtime gates
Env-var provenance forgery can mark a dirty tree clean
Existing plan/apply tests regress

Spec sources read

Forgejo issue #194
docs/forgejo-agent-operations.md
control-plane/platformctl/apply.py
control-plane/platformctl/plan.py
control-plane/platformctl/tests/test_apply_phase3.py
control-plane/platformctl/tests/test_plan_phase3.py
control-plane/platformctl/tests/test_apply_env_file.py

Validation

PYTHONPATH=control-plane control-plane/.venv/bin/python -m pytest control-plane/platformctl/tests/test_plan_phase3.py control-plane/platformctl/tests/test_apply_phase3.py control-plane/platformctl/tests/test_apply_env_file.py — 101 passed
PYTHONPATH=control-plane control-plane/.venv/bin/python -m platformctl.cli validate all --json — passed

Closes #194

Canary status: missing — fire canary 3+3 manually before merge ## Canary Context Pack ### Product story The apply pipeline should fail closed under hostile provenance inputs, not only under happy-path malformed values. ### What changed Added adversarial coverage for corrupted plan JSON, type-confused plan payloads, forged clean env state, type-confused source SHA values, and fake manifest git roots. ### Why it changed Issue #194 closes the redaction/provenance hardening segment after #199, #200, #206, #192, and #193. These tests document the threat model before moving to plan-state/runtime-safety tasks. ### Files touched - `control-plane/platformctl/apply.py` - `control-plane/platformctl/tests/test_apply_phase3.py` - `control-plane/platformctl/tests/test_plan_phase3.py` ### Relevant context - Issue #194: PROVENANCE-ADVERSARIAL-TESTS-01 - #199/#200/#206 redaction chain merged - #192/#193 provenance primitives merged ### Runtime evidence No runtime mutation. Local validation only. ### Known constraints This is intentionally test-heavy. The only code behavior change is fail-closed handling for invalid/non-object plan JSON before approval lookup. ### Explicit out-of-scope - No plan-state changes (#196/#197 next) - No runtime apply or service restart - No destructive operations ### Requested decision Approve merge if tests and Patchwarden checks are green. ### Merge blockers - Corrupted/type-confused plan JSON reaches approval/runtime gates - Env-var provenance forgery can mark a dirty tree clean - Existing plan/apply tests regress ## Spec sources read - Forgejo issue #194 - `docs/forgejo-agent-operations.md` - `control-plane/platformctl/apply.py` - `control-plane/platformctl/plan.py` - `control-plane/platformctl/tests/test_apply_phase3.py` - `control-plane/platformctl/tests/test_plan_phase3.py` - `control-plane/platformctl/tests/test_apply_env_file.py` ## Validation - `PYTHONPATH=control-plane control-plane/.venv/bin/python -m pytest control-plane/platformctl/tests/test_plan_phase3.py control-plane/platformctl/tests/test_apply_phase3.py control-plane/platformctl/tests/test_apply_env_file.py` — 101 passed - `PYTHONPATH=control-plane control-plane/.venv/bin/python -m platformctl.cli validate all --json` — passed Closes #194

codex added 1 commit

2026-05-30 15:24:22 +02:00

test(platformctl): add provenance adversarial coverage

platformctl plan / auto-apply scope (pull_request) Successful in 27s

Details

python-ci / Python 3.11 (pull_request) Successful in 46s

Details

python-ci / Python 3.12 (pull_request) Successful in 45s

Details

python-ci / Python 3.13 (pull_request) Successful in 45s

Details

base-is-main / guard (pull_request) Successful in 1s

Details

patchwarden-pr-sanity / collect-diff (pull_request) Successful in 6s

Details

canary-required / canary (pull_request) Successful in 15s

Details

patchwarden-client-dry-run / dry-run (pull_request) Successful in 23s

Details

canary-required / collect-diff (pull_request) Successful in 5s

Details

patchwarden-client-dry-run / collect-diff (pull_request) Successful in 5s

Details

pyfallow / Pyfallow gate (control-plane) (pull_request) Successful in 24s

Details

patchwarden-pr-sanity / sanity (pull_request) Successful in 2m1s

Details

55f33d3f5e

codex added the

class/security-sensitive

label

2026-05-30 15:24:23 +02:00

codex commented

2026-05-30 15:30:36 +02:00

Author

Collaborator

Patchwarden PR sanity

Status: advisory_findings
PR: 641
Commit: 55f33d3f5ee9157cee948173d4b168e0be5a0bb8
Security-sensitive label: present
Authority: advisory model review plus deterministic blockers only
3+3 canary: still alive; this does not replace it

Deterministic findings

info sensitive-path-touched Sensitive path touched — control-plane/platformctl/apply.py
- Evidence: control-plane/platformctl/apply.py
- Next: Route through the existing 3+3/risk-tier process; model review remains advisory.

Model reviewers

`global-glm` / `glm-5.1:cloud`

Status: ok
Verdict: OK
Findings: none

`global-deepseek` / `deepseek-v4-pro:cloud`

Status: ok
Verdict: OK
Findings: none

`redteam` / `kimi-k2.6:cloud`

Status: ok
Verdict: NOT_OK
high UnicodeDecodeError bypasses fail-closed plan parsing
- Evidence: control-plane/platformctl/apply.py: the new try/except wraps plan = json.loads(plan_path.read_text())but only catchesjson.JSONDecodeError. If the plan file contains non-UTF-8 bytes, read_text()raisesUnicodeDecodeErrorbeforejs`
- Next: Catch UnicodeDecodeError (and ideally OSError) alongside json.JSONDecodeError, or read bytes and decode explicitly, ensuring any plan-read failure returns the same structured error dict before the approval gate.

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 - Status: `advisory_findings` - PR: `641` - Commit: `55f33d3f5ee9157cee948173d4b168e0be5a0bb8` - Security-sensitive label: `present` - Authority: advisory model review plus deterministic blockers only - 3+3 canary: still alive; this does not replace it ## Deterministic findings - **`info` `sensitive-path-touched`** Sensitive path touched — `control-plane/platformctl/apply.py` - Evidence: `control-plane/platformctl/apply.py` - Next: Route through the existing 3+3/risk-tier process; model review remains advisory. ## Model reviewers ### `global-glm` / `glm-5.1:cloud` - Status: `ok` - Verdict: `OK` - Findings: none ### `global-deepseek` / `deepseek-v4-pro:cloud` - Status: `ok` - Verdict: `OK` - Findings: none ### `redteam` / `kimi-k2.6:cloud` - Status: `ok` - Verdict: `NOT_OK` - **`high`** UnicodeDecodeError bypasses fail-closed plan parsing - Evidence: `control-plane/platformctl/apply.py: the new try/except wraps `plan = json.loads(plan_path.read_text())` but only catches `json.JSONDecodeError`. If the plan file contains non-UTF-8 bytes, `read_text()` raises `UnicodeDecodeError` before `js` - Next: Catch `UnicodeDecodeError` (and ideally `OSError`) alongside `json.JSONDecodeError`, or read bytes and decode explicitly, ensuring any plan-read failure returns the same structured error dict before the approval gate. ## 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.

pdurlej approved these changes

2026-05-30 15:34:52 +02:00

pdurlej left a comment

Owner-authorized admin approval after green checks. Scope: #194 adds hostile provenance input tests and fail-closed invalid plan JSON handling; no runtime mutation and no secrets printed.