@levnikolaevich/hex-line-mcp 1.4.0 → 1.6.0

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/output-style.md CHANGED
@@ -1,33 +1,33 @@
1
1
  ---
2
2
  name: hex-line
3
- description: hex-line MCP tool preferences + explanatory coding style with insights
3
+ description: hex-line MCP tool preferences with compact coding style
4
4
  keep-coding-instructions: true
5
5
  ---
6
6
 
7
7
  # MCP Tool Preferences
8
8
 
9
- **PREFER** hex-line MCP for code files — hash-annotated reads enable safe edits:
9
+ **PREFER** hex-line MCP for code files. Hash-annotated reads and verified edits keep context cheap and safe.
10
10
 
11
11
  | Instead of | Use | Why |
12
12
  |-----------|-----|-----|
13
13
  | Read | `mcp__hex-line__read_file` | Hash-annotated, revision-aware |
14
14
  | Edit | `mcp__hex-line__edit_file` | Hash-verified anchors + conservative auto-rebase |
15
15
  | Write | `mcp__hex-line__write_file` | No prior Read needed |
16
- | Grep | `mcp__hex-line__grep_search` | Hash-annotated matches |
16
+ | Grep | `mcp__hex-line__grep_search` | Edit-ready matches |
17
17
  | Edit (text rename) | `mcp__hex-line__bulk_replace` | Multi-file text rename/refactor |
18
18
  | Bash `find`/`tree` | `mcp__hex-line__directory_tree` | Pattern search, gitignore-aware |
19
19
 
20
20
  ## Efficient File Reading
21
21
 
22
- For UNFAMILIAR code files >100 lines, PREFER:
23
- 1. `outline` first (code files only — not .md/.json/.yaml)
24
- 2. `read_file` with offset/limit for the specific section you need
25
- 3. Batch: `paths` array reads multiple files in one call
22
+ For unfamiliar code files >100 lines, prefer:
23
+ 1. `outline` first
24
+ 2. `read_file` with `offset`/`limit` or `ranges`
25
+ 3. `paths` or `ranges` when batching several targets
26
26
 
27
- Avoid reading a large file in full outline+targeted read saves 75% tokens.
27
+ Avoid reading a large file in full. Prefer compact, targeted reads.
28
28
 
29
29
  Bash OK for: npm/node/git/docker/curl, pipes, compound commands.
30
- **Built-in OK for:** images, PDFs, notebooks, Glob (always), `.claude/settings.json` and `.claude/settings.local.json`.
30
+ **Built-in OK for:** images, PDFs, notebooks, Glob (always), `.claude/settings.json`, `.claude/settings.local.json`.
31
31
 
32
32
  ## Edit Workflow
33
33
 
@@ -35,24 +35,26 @@ Prefer:
35
35
  1. collect all known hunks for one file
36
36
  2. send one `edit_file` call with batched edits
37
37
  3. carry `revision` from `read_file` into `base_revision` on follow-up edits
38
- 4. edit types: `set_line` (1 line), `replace_lines` (range + checksum), `insert_after`, `replace_between` (large blocks)
39
- 5. use `verify` before rereading a file after staleness
38
+ 4. use `set_line`, `replace_lines`, `insert_after`, `replace_between` based on scope
39
+ 5. use `verify` before rereading after staleness
40
+
41
+ Post-edit output uses `block: post_edit` with checksum — use it directly for follow-up edits or verify.
40
42
 
41
43
  Avoid:
42
44
  - chained same-file `edit_file` calls when all edits are already known
43
45
  - full-file rewrites for local changes
44
46
  - using `bulk_replace` for structural block rewrites
45
47
 
46
- # Explanatory Style
47
-
48
- Provide educational insights about the codebase alongside task completion. When providing insights, you may exceed typical length constraints, but remain focused and relevant.
49
-
50
- ## Insights
48
+ # Response Style
51
49
 
52
- Before and after writing code, provide brief educational explanations about implementation choices using:
50
+ Keep responses compact and operational. Explain only what is needed to complete the task or justify a non-obvious decision.
53
51
 
54
- "`\u2736 Insight \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500`
55
- [2-3 key educational points]
56
- `\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500`"
52
+ Prefer:
53
+ - short progress updates
54
+ - direct tool calls without discovery chatter
55
+ - concise summaries of edits and verification
57
56
 
58
- Focus on insights specific to the codebase or the code just written, not general programming concepts.
57
+ Avoid:
58
+ - mandatory educational blocks
59
+ - long prose around tool usage
60
+ - repeating obvious implementation details
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@levnikolaevich/hex-line-mcp",
3
- "version": "1.4.0",
3
+ "version": "1.6.0",
4
4
  "mcpName": "io.github.levnikolaevich/hex-line-mcp",
5
5
  "type": "module",
6
6
  "description": "Hash-verified file editing MCP + token efficiency hook for AI coding agents. 11 tools: read, edit, write, grep, outline, verify, directory_tree, file_info, setup_hooks, changes, bulk_replace.",
@@ -20,9 +20,9 @@
20
20
  "lint": "eslint .",
21
21
  "lint:fix": "eslint . --fix",
22
22
  "test": "node --test test/*.mjs",
23
- "benchmark": "node benchmark/index.mjs",
24
- "benchmark:diagnostic": "node benchmark/index.mjs --diagnostics",
25
- "benchmark:diagnostic:graph": "node benchmark/index.mjs --diagnostics --with-graph",
23
+ "scenarios": "node scenarios/index.mjs",
24
+ "scenarios:diagnostic": "node scenarios/index.mjs --diagnostics",
25
+ "scenarios:diagnostic:graph": "node scenarios/index.mjs --diagnostics --with-graph",
26
26
  "check": "node --check server.mjs && node --check hook.mjs"
27
27
  },
28
28
  "_dep_notes": {