mirror of
https://github.com/norandom/OpenUltraSAST.git
synced 2026-10-16 03:49:26 +00:00
docs(src): refresh comments and docstrings
Comment-only changes outside learn/, plane/ and cli.py: - model/pipeline.py: the `judge` its docstring contrasts with is the deleted model/judge.py, now named as such. - fusion.py: the panels are deterministic and only review and vote; the "critique and revise" step belonged to the retired LLM panels. - mcp.py: hunter tools stay inside the scan's stages, not "harness stages". - fusion, mcp, gate, redaction, skills: their "Phase N" tags now say which plan they cite (the retired openrouter-sast-harness). Requirement citations to retired specs stay: they explain where the code came from. No behaviour change; the byte-identical benchmark golden passes. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01EmM1j46qgKMh3zqD1QMwEo
This commit is contained in:
co-authored by
Claude Opus 5.5
parent
80eb0f2a36
commit
82c8a4d341
@@ -1,11 +1,11 @@
|
||||
"""Fusion adjudication — OpenUltraCode-style two-panel deepening (Phase 13).
|
||||
"""Fusion adjudication — OpenUltraCode-style two-panel deepening (Phase 13 of the retired openrouter-sast-harness plan).
|
||||
|
||||
Fusion is the deepening mechanism for findings that need more reasoning than the
|
||||
ranker → hunter → verifier → mapping loop provides: critical/high severity, verifier
|
||||
disagreement, static-vs-semantic evidence conflict, findings that gate a risky fix or
|
||||
disclosure, or an explicit high-assurance request. Two panels independently review the
|
||||
bounded evidence (one steel-manning the vulnerability case, one the false-positive
|
||||
case), critique and revise, vote, and a decider issues the final disposition.
|
||||
case) and vote, and a decider issues the final disposition.
|
||||
|
||||
The panels and the decider are deterministic and auditable. Every fused finding receives
|
||||
exactly one of the five dispositions, and the decision discloses panel roles, votes and the
|
||||
|
||||
@@ -1,4 +1,4 @@
|
||||
"""Detection-quality gate — the hard feasibility constraint (Phase 4 task 8).
|
||||
"""Detection-quality gate — the hard feasibility constraint (Phase 4 task 8 of the retired openrouter-sast-harness plan).
|
||||
|
||||
Aggregate recall and false-positive rate over the benchmark corpus must clear the
|
||||
project goal: recall >= 90% AND false-positive rate < 10%. This gate is computed
|
||||
|
||||
@@ -1,8 +1,8 @@
|
||||
"""Narrow MCP server for OpenCode integration (Phase 12).
|
||||
"""Narrow MCP server for OpenCode integration (Phase 12 of the retired openrouter-sast-harness plan).
|
||||
|
||||
Exposes only stable, project-level operations over MCP stdio (newline-delimited
|
||||
JSON-RPC 2.0), implemented with the standard library only. No tool exposes arbitrary
|
||||
shell, Docker, or internal hunter tools — those stay inside controlled harness stages.
|
||||
shell, Docker, or internal hunter tools — those stay inside the scan's own stages.
|
||||
The pure :meth:`McpServer.handle` dispatch is separated from the :func:`serve` stdio
|
||||
loop so the surface is unit-testable without a live client.
|
||||
"""
|
||||
|
||||
@@ -4,8 +4,8 @@ Everything before this module built an *arbiter* — something that judges a flo
|
||||
found on its own. Nothing proposed candidates to it, which is why the arbiter's recall was never comparable to
|
||||
a detector's: it could only report what it happened to find unaided.
|
||||
|
||||
This is the wiring the architecture was designed around, and it turns on a distinction that `judge` alone
|
||||
conflated:
|
||||
This is the wiring the architecture was designed around, and it turns on a distinction that the first judge
|
||||
(`model/judge.py`, deleted 2026-10-02 because this module replaced it) conflated:
|
||||
|
||||
the model CONTRADICTS it has coverage at this site and says no — a sanitizer lies on the path, or the
|
||||
sink is written in its declared-safe shape. The claim is dropped (Req 8.2).
|
||||
|
||||
@@ -1,4 +1,4 @@
|
||||
"""Secret redaction for traces and reports (Phase 15 hardening).
|
||||
"""Secret redaction for traces and reports (Phase 15 hardening of the retired openrouter-sast-harness plan).
|
||||
|
||||
Scan artifacts (trace events, reports) can quote source code that contains live
|
||||
credentials. This module masks well-known secret shapes before they are persisted, so
|
||||
|
||||
@@ -1,4 +1,4 @@
|
||||
"""Trail of Bits skill index + router (Phase 14).
|
||||
"""Trail of Bits skill index + router (Phase 14 of the retired openrouter-sast-harness plan).
|
||||
|
||||
Security skills are scoped expertise, not global prompt bloat. Each skill is a small
|
||||
descriptor with routing metadata (languages, tags, vulnerability classes, stages) and
|
||||
|
||||
Reference in New Issue
Block a user