audit-tools 0.51.3 → 0.51.4

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "audit-tools",
3
- "version": "0.51.3",
3
+ "version": "0.51.4",
4
4
  "private": false,
5
5
  "license": "ISC",
6
6
  "description": "Provider-neutral code auditing and remediation workflows for arbitrary repositories.",
@@ -36,18 +36,28 @@ Ask for exactly one step:
36
36
  audit-code next-step
37
37
  ```
38
38
 
39
+ Immediately after this first response, before following its workload prompt or
40
+ calling `next-step` again, perform a host-side structural-capability preflight.
41
+ Confirm that working tools can inspect the structural graph/relationships and
42
+ source structure required for the audit. An installed tool name or an unavailable
43
+ server is not evidence of capability. This check belongs to the host; do not add
44
+ provider, routing, model, or machine-capability fields to audit-tools.
45
+
46
+ If that capability is unavailable, state the concrete limitation and explicitly
47
+ say the audit is degraded/non-comprehensive. Stop before critical-flow analysis,
48
+ charter work, contract or conceptual review, per-file review, or writing semantic
49
+ results. Do not defer the notice until a report is written or silently substitute
50
+ source-only review for a comprehensive request. A user-chosen quick/shallow audit
51
+ may proceed with the stated limitation and the reflection required below.
52
+
39
53
  Read the returned JSON only far enough to find `prompt_path`, then read and
40
54
  follow only that prompt. Do not inspect workload, result, schema, command
41
55
  catalog, or state files unless the current prompt directs you to them.
42
56
 
43
57
  When the prompt emits semantic review items, assign them with the host's native
44
58
  subagent facilities when available. Do not send provider, model, quota,
45
- context-window, routing, or launch configuration to audit-tools. Before treating
46
- the run as comprehensive, perform a host-side structural-capability preflight:
47
- confirm the host can inspect the structural graph/relationships and source
48
- structure required by the workload. This is a host capability check, not an
49
- audit-tools MCP/provider or lane-selection step. If capability is degraded but
50
- you proceed, record one reserved AgentReflection with task_id exactly
59
+ context-window, routing, or launch configuration to audit-tools. If the user has
60
+ chosen a degraded quick/shallow run, record one reserved AgentReflection with task_id exactly
51
61
  `audit-capability-preflight`. Every reflection requires all three of `task_id`,
52
62
  `instruction_clarity` (one of `clear`, `mostly_clear`, `ambiguous`, `unclear`)
53
63
  and `severity`; a line missing any of the three is discarded whole. Use severity