@yeison.restrepo.r/code-conductor 1.31.0 → 1.31.2

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.
@@ -14,6 +14,8 @@ Identify and extract:
14
14
  - Add a timestamped section: `## Checkpoint [YYYY-MM-DD HH:MM]`
15
15
  - List decisions, conventions, and debt from this session
16
16
 
17
+ **Staging that update, where the workflow commits it:** stage by tracked-ness, with an explicit path. A `project.md` that is already tracked takes `git add -u ".claude/memory/project.md"`; one that is not yet tracked takes plain `git add ".claude/memory/project.md"`, or `git add -f` where the project ignores `.claude/`. **Never an unconditional `-u` here.** This command is global and also runs in freshly scaffolded projects, which do not ignore `.claude/` and whose `project.md` is not yet tracked: there `-u` exits 128 and would halt the checkpoint before its DB tail, trading a defect those projects never had for one they would.
18
+
17
19
  **Update `.claude/memory/personal.md`** (local only):
18
20
  - Add any developer preferences observed this session
19
21
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@yeison.restrepo.r/code-conductor",
3
- "version": "1.31.0",
3
+ "version": "1.31.2",
4
4
  "description": "A spec-first, token-efficient Claude Code configuration and installer CLI.",
5
5
  "repository": {
6
6
  "type": "git",
@@ -98,6 +98,13 @@ Each step must include:
98
98
  order, so a staging step placed above its edit commits the file's previous content while
99
99
  every checkbox still reports `[X]`, and the divergence surfaces only at whatever later task
100
100
  asserts on that file's expected state, pointing at the wrong task.
101
+ - **Staging form within a step.** Where a staged path sits under a directory the repository
102
+ ignores, plain `git add <path>` exits 1 even when a tracked file staged correctly, because
103
+ the pathspec matched the ignored ancestor. Generate `git add -u <path>` when the file is
104
+ already tracked, `git add -f <path>` when it is new and under an ignored directory, and
105
+ plain `git add <path>` otherwise. Always generate an explicit path: bare `git add -u`
106
+ stages every modified tracked file in the repository. The branch is on tracked-ness, never
107
+ on habit, because `-u` exits 128 on an untracked path whether or not an ignore rule exists.
101
108
 
102
109
  ## Test List
103
110
  - [ ] Unit tests for [unit]
@@ -125,8 +132,8 @@ instruction in step 7.
125
132
 
126
133
  **Precondition on Task 0.** This gate runs to completion — the warning, the offer, and
127
134
  either the confirmed `git switch` or an explicit decline —
128
- **before any step of the approved plan's Task 0 executes**, including its `git add` /
129
- `git add -f` and its `git commit`. Its position in this file is not the guarantee; the
135
+ **before any step of the approved plan's Task 0 executes**, including its `git add`,
136
+ `git add -u` or `git add -f` and its `git commit`. Its position in this file is not the guarantee; the
130
137
  execution order is.
131
138
  Under the plan-commit ritual, Task 0 runs at approval, which is exactly when a plan commit
132
139
  lands on the wrong branch.
@@ -195,11 +195,14 @@ function g3Scan(mode, input) {
195
195
 
196
196
  // Each check returns true when it FIRES (the authority's shell functions returned 1).
197
197
  //
198
- // A quirk worth naming, because it looks like a bug and is not: the authority walks
199
- // with `after="${rest:mlen}"`, slicing by the match LENGTH from position 0 rather
200
- // than from the match index. For `ls; cat *.ts` the match is `; cat ` and bash's
201
- // `after` is `at *.ts`, not `*.ts`. P4 and P7 reproduce that exactly. "Fixing" it
202
- // would make the port disagree with the corpus.
198
+ // The walk's contract, corrected in both subjects under [BUG-041]: `after` is the
199
+ // text that FOLLOWS the match, sliced from `m.index + m[0].length`. The original form
200
+ // sliced by the match LENGTH from position 0, so for a match at index N it shaved N
201
+ // bytes of real command off the front and re-scanned, walking leftward until the cut
202
+ // landed inside a quoted region; g3Scan's fail-closed clause then reported a glob that
203
+ // was never in the command. It cost 24 of the 47 denials measured in one session. The
204
+ // authority carries the same correction, recorded as its third sanctioned exception,
205
+ // so the corpus still arbitrates both subjects rather than one.
203
206
 
204
207
  function g3P1FindDepth(s) {
205
208
  if (!reFind.test(s)) return false;
@@ -241,7 +244,7 @@ function g3GlobWalk(s, re) {
241
244
  for (;;) {
242
245
  const m = re.exec(rest);
243
246
  if (!m) return false;
244
- const after = rest.slice(m[0].length); // authority quirk: slice by length, not index
247
+ const after = rest.slice(m.index + m[0].length);
245
248
  if (g3Scan('glob', after).glob) return true;
246
249
  rest = after;
247
250
  if (rest === '') return false;