sphica 0.6.41 → 0.6.43

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": "sphica",
3
- "version": "0.6.41",
3
+ "version": "0.6.43",
4
4
  "description": "Records Claude Code and Codex sessions on your machine and keeps past implementation and decisions, with their sources, for your agent to find.",
5
5
  "license": "MIT",
6
6
  "type": "module",
@@ -32,7 +32,7 @@ Pass the diff under review and the root of the repository under review (`cwd`) t
32
32
  |---|---|---|
33
33
  | The tool call fails, or it says "Decision lane: not checked" | MCP does not connect, the database is unreachable, or the project is not registered | **`blocked_unknown`** + the reason it gave |
34
34
  | "Decision lane: checked" with no record | No active record applies | Continue to Step 3; 0 records may be treated as a **grounded negative** |
35
- | "Decision lane: checked" with records | One batch of at most 50 records, `Batch k of n`, and a `selection`. Each record says why it applies (anchored to a changed path, or an added line names an option it rejected); an AI's decision is marked `decided by an AI` | Continue |
35
+ | "Decision lane: checked" with records | One batch of at most 50 records, `Batch k of n`, and a `selection`. Each record says why it applies (anchored to a changed path, or an added line names an option it rejected), cut short with `…` when long; an AI's decision is marked `decided by an AI`. A batch whose keys are too long to show lists `u<id>` alone and says so | Continue |
36
36
 
37
37
  **When returning `blocked_unknown`, state concretely what was missing.** Silently returning 0 results makes the caller read it as "no findings".
38
38
  **This step is deterministic and can claim coverage.** It selects only active records; candidates and superseded records never apply.
@@ -45,8 +45,10 @@ Changes to the working tree between batches (which code locations still exist) a
45
45
 
46
46
  ## Step 2 — Read every selected record
47
47
 
48
- `read([keys], cwd)` returns each record's text, its options, the exact words cited as evidence and adoption with who said them, what it
49
- superseded or conflicts with, and each code location checked in the working tree now. **Judge from this body, never from the key or the one line.**
48
+ Read them by the `u<id>` each line gives: `read(["u12", "u40"], cwd)`. A batch whose keys are too long to show names its records by
49
+ `u<id>` alone, and read shows each one's whole key, which its finding's `unit` takes. read returns each record's text, its options, the
50
+ exact words cited as evidence and adoption with who said them, what it superseded or conflicts with, and each code location checked in
51
+ the working tree now. **Judge from this body, never from the key or the one line**, which is cut short (`…`) to keep the batch whole.
50
52
  When a reply says some refs were not read or a record continues, call `read` again with exactly what it names until nothing is left.
51
53
 
52
54
  ## Step 3 — Search by the approach's meaning