mantis-report

What the source does

• FilesystemNone observedNo reads or writes in the source
• NetworkNone observedNo outbound calls in the source
• ShellNone observedNo commands run in the source
• DependenciesPinnedNothing floats
• SecretsNone foundNo keys in source

Five audit layers

Gates applied: no behavioural pass.

Audit 0.4.1 · source sha 9702f6eef495 · 29 Oct 2026Report an issue or request a re-scan

• Publishergoogle GitHub organizationOwner of the repository this skill is read from
• Source foldergoogle/mantis/skills/mantis-reportThe exact folder we audited
• LicenceApache-2.0As the repository states it
• Commit pinned for this verdict9702f6eef495A new audit runs when the source changes
• Repository statusNot archivedLast push 3 days ago
• Sandboxed runNot runSkills are read, not executed; layer L4 is skipped

Its description: Generates a human-readable security review packet compiled from confirmed findings and exploit chains. Use at the end of a review cycle to produce stakeholder-facing documentation. Don't use for auditing code or…

---
name: mantis-report
description: >-
  Generates a human-readable security review packet compiled from confirmed findings and exploit chains.
  Use at the end of a review cycle to produce stakeholder-facing documentation.
  Don't use for auditing code or verifying patches directly.
---

# Reporter (/mantis-report)

## System Goal

Security Reporting Expert. Synthesizes complex, technical finding logs into a
high-quality, human-readable review packet for developers and stakeholders.

## Command Definition

- **Command:** `/mantis-report`
- **Description:** Generates a human-readable security review packet containing
  confirmed findings and exploit chains.

## Input/Output Contract

- **Reads**:
  - `workspace/findings/*.json` (all active finding files for this pass).
  - `workspace/archive/findings_pass_*/*.json` (archived findings from prior
    passes) and legacy `workspace/archive/loop*_findings/*.json`. The report is
    a campaign-wide view: this pass's full findings plus carried-forward
    findings that prior passes stopped retrying (e.g. hit the retry cap) so they
    do not vanish from the report.
  - `workspace/.mantis_state.json` (to track current loop pass, and to read
    `vcs_info` and `active_snapshot` {`root`, `snapshot_id`, `snapshot_pinned`}
    for provenance).
  - Per-finding snapshot provenance fields, all OPTIONAL: `discovery_commit`,
    `repro_snapshot_id`, `patch_base_snapshot`. When any is absent/empty it is
    rendered as "(not recorded)" — never a reason to drop a finding.
- **Writes**:
  - `workspace/report/review_packet_pass_<N>_<snapshot_tag>.md` (pass- and
    snapshot-tagged markdown report). Falls back to the unsuffixed
    `review_packet_pass_<N>.md` on a legacy pass with no recorded snapshot.
  - Updates copy/symlink at `workspace/report/review_packet-latest.md`.
- **Preconditions**:
  - Calibrated and reproduced findings exist in `workspace/findings/` or
    `workspace/archive/`. The report is a campaign-wide view: it reports the
    current state of every unresolved finding discovered in this or any prior
    pass, de-duplicated to each finding's latest state. A confirmed-but-unfixed
    finding that plan stopped carrying back (e.g. it hit the 2-attempt retry
    cap) does NOT vanish — it appears here at its most recent archived state.
- **Idempotency Guarantee**:
  - Writes to pass-and-snapshot-tagged files. In-place overwrite of
    `review_packet-latest.md`. Re-running the SAME pass on the SAME snapshot
    updates the same tagged file. Re-running the same pass number on a DIFFERENT
    snapshot writes a DISTINCT file (the `<snapshot_tag>` suffix prevents
    cross-snapshot overwrite). Legacy passes with no recorded snapshot keep the
    unsuffixed `review_packet_pass_<N>.md` name and overwrite in place, exactly
    as before.

## Instructions

**Step 0 — Locator Resolution.**

```
LOCATOR RESOLUTION (before reading ANY target code or artifact):
0. ROLE: If this skill NEVER reads target source (report, calibrate, reflect),
   you are a FINDINGS-ONLY stage: skip steps 2-6; still read active_snapshot from
   state for provenance/annotation; NEVER stop merely because a code root is unset.
1. Determine CODE_ROOT, in this priority order:
   a. If --target_root is passed on THIS invocation, CODE_ROOT = --target_root.
      It is AUTHORITATIVE and OVERRIDES SNAPSHOT_ROOT and the state fallback
      (used when a caller hands you a prepared tree, e.g. a patched shadow).
   b. Else if --snapshot_root (or SNAPSHOT_ROOT) is passed, use it.
   c. Else read state_root/workspace/.mantis_state.json (state_root from
      --state_root if passed, else ./workspace/... relative to the current dir)
      -> active_snapshot.root / .snapshot_id / .snapshot_pinned.
   d. Else (no arg AND no readable active_snapshot): CODE_ROOT = current directory,
      treat snapshot_pinned = false (MODE-OFF). Do NOT stop.
2. SENTINEL CHECK (only if snapshot_pinned is true AND you did NOT take path 1a):
   verify CODE_ROOT/.mantis_snapshot_id exists and equals SNAPSHOT_ID. If missing
   or different -> STOP "snapshot sentinel mismatch". (A --target_root tree (1a) is
   deliberately mutated and is sentinel-EXEMPT.)
3. PATH FIELDS:
   - SNAPSHOT-RELATIVE (read under CODE_ROOT): code_paths entries; plan target_files
     that are file paths. Strip ONLY a trailing ":<digits>". A code_paths entry
     containing "://" is a URL/endpoint, NOT a file read. A code_paths entry that is
     NOT of the form <existing-path>:<integer> is a non-source LOCATOR
     (symbol/offset/endpoint): only check that the artifact/symbol exists; skip ALL
     line-range and line-existence logic.
   - STATE-RELATIVE (read/write under state_root/workspace, NEVER prefix CODE_ROOT):
     kb_references, repro_file_path, reattack_file_path, helper scripts, report
     files, and all state/findings JSON.
4. Never WRITE under CODE_ROOT when snapshot_pinned is true. Any command that
   compiles, generates, or writes artifacts MUST run in a PRIVATE SHADOW copy
   (mktemp -d from CODE_ROOT), never with cwd=CODE_ROOT. Read-only inspection may
   cd into CODE_ROOT.
5. VCS-METADATA CARVE-OUT: history-log extraction and any VCS diff/blame command
   run in the LIVE repository root (which still has .git/.hg/.repo), NOT CODE_ROOT
   (the snapshot copy strips VCS metadata). Do NOT stop merely because CODE_ROOT
   lacks .git/.hg/.repo.
6. Every shell command uses ABSOLUTE paths and sets its own working directory on
   that call. Do NOT assume the working directory persists between calls.
```

Then the following FINDINGS-ONLY notes apply to the Reporter:

- The Reporter is a FINDINGS-ONLY stage (Block A, ROLE step 0): it SKIPS locator
  steps 2–6, does NOT require or resolve a CODE_ROOT, and NEVER stops because a
  code root or sentinel is unset.
- It STILL reads `active_snapshot` (`root`, `snapshot_id`, `snapshot_pinned`)
  and `vcs_info` from `workspace/.mantis_state.json` for provenance — used to
  build the header, the top b

Excerpt, first 6,000 characters of SKILL.md, verbatim from the audited commit. It is text from someone else’s repository, quoted so you can judge it before your agent reads it.

git clone https://github.com/google/mantis
git -C mantis checkout 9702f6eef495

Works in

1. 2026-10-29
   ✓SAFE B 89 / 100

   first audit · source 9702f6eef495

Questions about mantis-report

• github.com/google/mantis/tree/main/skills/mantis-report — the audited source at commit 9702f6eef495 and every finding aboveObserved · 29 Oct 2026

Spotted something wrong? Report a correction · Methodology