viepilot 2.50.1 → 3.7.3

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.
Files changed (49) hide show
  1. package/CHANGELOG.md +214 -0
  2. package/README.md +17 -17
  3. package/bin/viepilot.cjs +1 -0
  4. package/bin/vp-tools.cjs +123 -1
  5. package/docs/brainstorm/session-2026-05-22.md +472 -0
  6. package/docs/dev/agents.md +51 -41
  7. package/docs/dev/architecture.md +26 -0
  8. package/docs/skills-reference.md +96 -0
  9. package/lib/adapter-context.cjs +294 -0
  10. package/lib/adapters/antigravity.cjs +8 -2
  11. package/lib/adapters/claude-code.cjs +4 -0
  12. package/lib/audit/browser-runner.cjs +102 -0
  13. package/lib/intake/adapters/browser.cjs +58 -0
  14. package/lib/intake/adapters/excel-m365.cjs +54 -6
  15. package/lib/intake/auto-intake.cjs +194 -0
  16. package/lib/intake/classifier.cjs +22 -4
  17. package/lib/intake/manifest.cjs +81 -0
  18. package/lib/intake/triage-ux.cjs +10 -2
  19. package/lib/intake/validator.cjs +97 -0
  20. package/lib/intake/writeback.cjs +169 -3
  21. package/lib/request/url-enricher.cjs +69 -0
  22. package/lib/viepilot-install.cjs +15 -0
  23. package/package.json +1 -1
  24. package/skills/vp-audit/SKILL.md +99 -3
  25. package/skills/vp-auto/SKILL.md +54 -4
  26. package/skills/vp-brainstorm/SKILL.md +69 -3
  27. package/skills/vp-crystallize/SKILL.md +52 -3
  28. package/skills/vp-debug/SKILL.md +52 -3
  29. package/skills/vp-design/SKILL.md +52 -3
  30. package/skills/vp-docs/SKILL.md +52 -3
  31. package/skills/vp-evolve/SKILL.md +52 -3
  32. package/skills/vp-info/SKILL.md +52 -3
  33. package/skills/vp-intake/SKILL.md +306 -7
  34. package/skills/vp-pause/SKILL.md +52 -3
  35. package/skills/vp-persona/SKILL.md +52 -3
  36. package/skills/vp-proposal/SKILL.md +52 -3
  37. package/skills/vp-request/SKILL.md +72 -3
  38. package/skills/vp-resume/SKILL.md +52 -3
  39. package/skills/vp-rollback/SKILL.md +52 -3
  40. package/skills/vp-skills/SKILL.md +52 -3
  41. package/skills/vp-status/SKILL.md +52 -3
  42. package/skills/vp-task/SKILL.md +52 -3
  43. package/skills/vp-ui-components/SKILL.md +52 -3
  44. package/skills/vp-update/SKILL.md +52 -3
  45. package/workflows/autonomous.md +268 -18
  46. package/workflows/brainstorm.md +222 -7
  47. package/workflows/crystallize.md +124 -6
  48. package/workflows/design.md +62 -1
  49. package/workflows/request.md +54 -8
@@ -53,7 +53,7 @@ Silent if command unavailable or errors.
53
53
  </persona_context>
54
54
 
55
55
 
56
- <cursor_skill_adapter>
56
+ <adapter id="claude-code">
57
57
  ## A. Skill Invocation
58
58
  - Skill được gọi khi user mention `vp-crystallize`, `/vp-crystallize`, hoặc "crystallize", "setup project"
59
59
  - Treat all user text after the skill mention as `{{VP_ARGS}}`
@@ -62,8 +62,57 @@ Silent if command unavailable or errors.
62
62
  Prompt user conversationally with numbered list options.
63
63
 
64
64
  ## C. Tool Usage
65
- Use Cursor tools: `Shell`, `ReadFile`, `Glob`, `rg`, `ApplyPatch`, `WebSearch`, `WebFetch`, `Subagent`
66
- </cursor_skill_adapter>
65
+ Use Claude Code tools: `Bash` (shell), `Read` (file), `Edit` + `Write` (file write/patch),
66
+ `Grep` (search), `Glob` (file patterns), `LS`, `WebSearch`, `WebFetch`,
67
+ `Agent` (spawn subagent — multi-level nesting supported)
68
+ Interactive: `AskUserQuestion` (deferred — preload via ToolSearch before first call)
69
+ </adapter>
70
+
71
+ <adapter id="cursor-agent">
72
+ ## A. Skill Invocation
73
+ Same trigger keywords as claude-code adapter.
74
+
75
+ ## C. Tool Usage
76
+ Use Cursor tools: `run_terminal_cmd` (shell), `read_file` (read), `edit_file` (write/edit),
77
+ `grep_search` (search), `web_search`, `codebase_search`, `list_dir`, `file_search`
78
+ Interactive: text list fallback (AskQuestion available in Plan Mode only; Agent Mode = text)
79
+ Subagent: `/multitask` (user command, single-level only — not a callable tool)
80
+ MCP limit: 40 tools
81
+ </adapter>
82
+
83
+ <adapter id="antigravity">
84
+ ## A. Skill Invocation
85
+ Same trigger keywords as claude-code adapter.
86
+ Skill discovery: LLM-driven (automatic, no slash command needed).
87
+
88
+ ## C. Tool Usage
89
+ Use Antigravity tools: `shell` (cmd), `file_read`, `file_write`, MCP plugins
90
+ Interactive: text fallback (TUI-based; no formal AskUserQuestion)
91
+ Skill path: `.agents/skills/<skill>/SKILL.md` (project) or `~/.gemini/antigravity/skills/` (global)
92
+ Note: Gemini CLI deprecated June 18, 2026 — use Antigravity CLI.
93
+ </adapter>
94
+
95
+ <adapter id="codex">
96
+ ## A. Skill Invocation
97
+ Same trigger keywords as claude-code adapter.
98
+
99
+ ## C. Tool Usage
100
+ Use Codex tools: `container.exec` (sandboxed shell), `apply_patch` (file write), `web_search`
101
+ Interactive: text fallback (TUI Tab/Enter injection)
102
+ Config: `~/.codex/config.toml`
103
+ </adapter>
104
+
105
+ <adapter id="copilot">
106
+ ## A. Skill Invocation
107
+ Same trigger keywords as claude-code adapter.
108
+ Discovery: User-driven (`@agent-name` in GitHub Copilot Chat).
109
+
110
+ ## C. Tool Usage
111
+ Use Copilot tools: `runCommands` (shell), `read`/`readfile` (read), `edit`/`editFiles` (write),
112
+ `code_search`, `find_references`
113
+ Interactive: `askQuestions` (main agent only — NOT available in subagents; VS Code issue #293745)
114
+ Skill path: `.github/agents/<name>.agent.md`
115
+ </adapter>
67
116
  <scope_policy>
68
117
  ## ViePilot Namespace Guard (BUG-004)
69
118
  - Default mode: only use and reference `vp-*` skills in ViePilot workflows.
@@ -53,7 +53,7 @@ Silent if command unavailable or errors.
53
53
  </persona_context>
54
54
 
55
55
 
56
- <cursor_skill_adapter>
56
+ <adapter id="claude-code">
57
57
  ## A. Skill Invocation
58
58
  - Skill được gọi khi user mention `vp-debug`, `/vp-debug`, "debug", "gỡ lỗi"
59
59
  - Treat all user text after the skill mention as `{{VP_ARGS}}`
@@ -62,8 +62,57 @@ Silent if command unavailable or errors.
62
62
  Prompt user conversationally. Guide through systematic debugging steps.
63
63
 
64
64
  ## C. Tool Usage
65
- Use Cursor tools: `Shell`, `ReadFile`, `Glob`, `rg`, `ApplyPatch`, `WebSearch`, `WebFetch`, `Subagent`
66
- </cursor_skill_adapter>
65
+ Use Claude Code tools: `Bash` (shell), `Read` (file), `Edit` + `Write` (file write/patch),
66
+ `Grep` (search), `Glob` (file patterns), `LS`, `WebSearch`, `WebFetch`,
67
+ `Agent` (spawn subagent — multi-level nesting supported)
68
+ Interactive: `AskUserQuestion` (deferred — preload via ToolSearch before first call)
69
+ </adapter>
70
+
71
+ <adapter id="cursor-agent">
72
+ ## A. Skill Invocation
73
+ Same trigger keywords as claude-code adapter.
74
+
75
+ ## C. Tool Usage
76
+ Use Cursor tools: `run_terminal_cmd` (shell), `read_file` (read), `edit_file` (write/edit),
77
+ `grep_search` (search), `web_search`, `codebase_search`, `list_dir`, `file_search`
78
+ Interactive: text list fallback (AskQuestion available in Plan Mode only; Agent Mode = text)
79
+ Subagent: `/multitask` (user command, single-level only — not a callable tool)
80
+ MCP limit: 40 tools
81
+ </adapter>
82
+
83
+ <adapter id="antigravity">
84
+ ## A. Skill Invocation
85
+ Same trigger keywords as claude-code adapter.
86
+ Skill discovery: LLM-driven (automatic, no slash command needed).
87
+
88
+ ## C. Tool Usage
89
+ Use Antigravity tools: `shell` (cmd), `file_read`, `file_write`, MCP plugins
90
+ Interactive: text fallback (TUI-based; no formal AskUserQuestion)
91
+ Skill path: `.agents/skills/<skill>/SKILL.md` (project) or `~/.gemini/antigravity/skills/` (global)
92
+ Note: Gemini CLI deprecated June 18, 2026 — use Antigravity CLI.
93
+ </adapter>
94
+
95
+ <adapter id="codex">
96
+ ## A. Skill Invocation
97
+ Same trigger keywords as claude-code adapter.
98
+
99
+ ## C. Tool Usage
100
+ Use Codex tools: `container.exec` (sandboxed shell), `apply_patch` (file write), `web_search`
101
+ Interactive: text fallback (TUI Tab/Enter injection)
102
+ Config: `~/.codex/config.toml`
103
+ </adapter>
104
+
105
+ <adapter id="copilot">
106
+ ## A. Skill Invocation
107
+ Same trigger keywords as claude-code adapter.
108
+ Discovery: User-driven (`@agent-name` in GitHub Copilot Chat).
109
+
110
+ ## C. Tool Usage
111
+ Use Copilot tools: `runCommands` (shell), `read`/`readfile` (read), `edit`/`editFiles` (write),
112
+ `code_search`, `find_references`
113
+ Interactive: `askQuestions` (main agent only — NOT available in subagents; VS Code issue #293745)
114
+ Skill path: `.github/agents/<name>.agent.md`
115
+ </adapter>
67
116
  <scope_policy>
68
117
  ## ViePilot Namespace Guard (BUG-004)
69
118
  - Default mode: only use and reference `vp-*` skills in ViePilot workflows.
@@ -46,7 +46,7 @@ Inject the output as `## User Persona` context before any task execution.
46
46
  Silent if command unavailable or errors.
47
47
  </persona_context>
48
48
 
49
- <cursor_skill_adapter>
49
+ <adapter id="claude-code">
50
50
  ## A. Skill Invocation
51
51
  - Skill is triggered when user mentions `vp-design`, `/vp-design`, "design system", "design.md", "sync tokens", "audit design"
52
52
  - Treat all user text after the skill mention as `{{VP_ARGS}}`
@@ -55,8 +55,57 @@ Silent if command unavailable or errors.
55
55
  Prompt user conversationally with numbered list options at control points.
56
56
 
57
57
  ## C. Tool Usage
58
- Use Cursor tools: `Shell`, `ReadFile`, `Glob`, `rg`, `ApplyPatch`, `WebSearch`, `WebFetch`, `Subagent`
59
- </cursor_skill_adapter>
58
+ Use Claude Code tools: `Bash` (shell), `Read` (file), `Edit` + `Write` (file write/patch),
59
+ `Grep` (search), `Glob` (file patterns), `LS`, `WebSearch`, `WebFetch`,
60
+ `Agent` (spawn subagent — multi-level nesting supported)
61
+ Interactive: `AskUserQuestion` (deferred — preload via ToolSearch before first call)
62
+ </adapter>
63
+
64
+ <adapter id="cursor-agent">
65
+ ## A. Skill Invocation
66
+ Same trigger keywords as claude-code adapter.
67
+
68
+ ## C. Tool Usage
69
+ Use Cursor tools: `run_terminal_cmd` (shell), `read_file` (read), `edit_file` (write/edit),
70
+ `grep_search` (search), `web_search`, `codebase_search`, `list_dir`, `file_search`
71
+ Interactive: text list fallback (AskQuestion available in Plan Mode only; Agent Mode = text)
72
+ Subagent: `/multitask` (user command, single-level only — not a callable tool)
73
+ MCP limit: 40 tools
74
+ </adapter>
75
+
76
+ <adapter id="antigravity">
77
+ ## A. Skill Invocation
78
+ Same trigger keywords as claude-code adapter.
79
+ Skill discovery: LLM-driven (automatic, no slash command needed).
80
+
81
+ ## C. Tool Usage
82
+ Use Antigravity tools: `shell` (cmd), `file_read`, `file_write`, MCP plugins
83
+ Interactive: text fallback (TUI-based; no formal AskUserQuestion)
84
+ Skill path: `.agents/skills/<skill>/SKILL.md` (project) or `~/.gemini/antigravity/skills/` (global)
85
+ Note: Gemini CLI deprecated June 18, 2026 — use Antigravity CLI.
86
+ </adapter>
87
+
88
+ <adapter id="codex">
89
+ ## A. Skill Invocation
90
+ Same trigger keywords as claude-code adapter.
91
+
92
+ ## C. Tool Usage
93
+ Use Codex tools: `container.exec` (sandboxed shell), `apply_patch` (file write), `web_search`
94
+ Interactive: text fallback (TUI Tab/Enter injection)
95
+ Config: `~/.codex/config.toml`
96
+ </adapter>
97
+
98
+ <adapter id="copilot">
99
+ ## A. Skill Invocation
100
+ Same trigger keywords as claude-code adapter.
101
+ Discovery: User-driven (`@agent-name` in GitHub Copilot Chat).
102
+
103
+ ## C. Tool Usage
104
+ Use Copilot tools: `runCommands` (shell), `read`/`readfile` (read), `edit`/`editFiles` (write),
105
+ `code_search`, `find_references`
106
+ Interactive: `askQuestions` (main agent only — NOT available in subagents; VS Code issue #293745)
107
+ Skill path: `.github/agents/<name>.agent.md`
108
+ </adapter>
60
109
 
61
110
  <scope_policy>
62
111
  ## ViePilot Namespace Guard (BUG-004)
@@ -53,7 +53,7 @@ Silent if command unavailable or errors.
53
53
  </persona_context>
54
54
 
55
55
 
56
- <cursor_skill_adapter>
56
+ <adapter id="claude-code">
57
57
  ## A. Skill Invocation
58
58
  - Skill được gọi khi user mention `vp-docs`, `/vp-docs`, "docs", "documentation", "tài liệu"
59
59
  - Treat all user text after the skill mention as `{{VP_ARGS}}`
@@ -62,8 +62,57 @@ Silent if command unavailable or errors.
62
62
  Prompt user conversationally with options.
63
63
 
64
64
  ## C. Tool Usage
65
- Use Cursor tools: `Shell`, `ReadFile`, `Glob`, `rg`, `ApplyPatch`, `WebSearch`, `WebFetch`, `Subagent`
66
- </cursor_skill_adapter>
65
+ Use Claude Code tools: `Bash` (shell), `Read` (file), `Edit` + `Write` (file write/patch),
66
+ `Grep` (search), `Glob` (file patterns), `LS`, `WebSearch`, `WebFetch`,
67
+ `Agent` (spawn subagent — multi-level nesting supported)
68
+ Interactive: `AskUserQuestion` (deferred — preload via ToolSearch before first call)
69
+ </adapter>
70
+
71
+ <adapter id="cursor-agent">
72
+ ## A. Skill Invocation
73
+ Same trigger keywords as claude-code adapter.
74
+
75
+ ## C. Tool Usage
76
+ Use Cursor tools: `run_terminal_cmd` (shell), `read_file` (read), `edit_file` (write/edit),
77
+ `grep_search` (search), `web_search`, `codebase_search`, `list_dir`, `file_search`
78
+ Interactive: text list fallback (AskQuestion available in Plan Mode only; Agent Mode = text)
79
+ Subagent: `/multitask` (user command, single-level only — not a callable tool)
80
+ MCP limit: 40 tools
81
+ </adapter>
82
+
83
+ <adapter id="antigravity">
84
+ ## A. Skill Invocation
85
+ Same trigger keywords as claude-code adapter.
86
+ Skill discovery: LLM-driven (automatic, no slash command needed).
87
+
88
+ ## C. Tool Usage
89
+ Use Antigravity tools: `shell` (cmd), `file_read`, `file_write`, MCP plugins
90
+ Interactive: text fallback (TUI-based; no formal AskUserQuestion)
91
+ Skill path: `.agents/skills/<skill>/SKILL.md` (project) or `~/.gemini/antigravity/skills/` (global)
92
+ Note: Gemini CLI deprecated June 18, 2026 — use Antigravity CLI.
93
+ </adapter>
94
+
95
+ <adapter id="codex">
96
+ ## A. Skill Invocation
97
+ Same trigger keywords as claude-code adapter.
98
+
99
+ ## C. Tool Usage
100
+ Use Codex tools: `container.exec` (sandboxed shell), `apply_patch` (file write), `web_search`
101
+ Interactive: text fallback (TUI Tab/Enter injection)
102
+ Config: `~/.codex/config.toml`
103
+ </adapter>
104
+
105
+ <adapter id="copilot">
106
+ ## A. Skill Invocation
107
+ Same trigger keywords as claude-code adapter.
108
+ Discovery: User-driven (`@agent-name` in GitHub Copilot Chat).
109
+
110
+ ## C. Tool Usage
111
+ Use Copilot tools: `runCommands` (shell), `read`/`readfile` (read), `edit`/`editFiles` (write),
112
+ `code_search`, `find_references`
113
+ Interactive: `askQuestions` (main agent only — NOT available in subagents; VS Code issue #293745)
114
+ Skill path: `.github/agents/<name>.agent.md`
115
+ </adapter>
67
116
  <scope_policy>
68
117
  ## ViePilot Namespace Guard (BUG-004)
69
118
  - Default mode: only use and reference `vp-*` skills in ViePilot workflows.
@@ -53,7 +53,7 @@ Silent if command unavailable or errors.
53
53
  </persona_context>
54
54
 
55
55
 
56
- <cursor_skill_adapter>
56
+ <adapter id="claude-code">
57
57
  ## A. Skill Invocation
58
58
  - Skill được gọi khi user mention `vp-evolve`, `/vp-evolve`, "evolve", "thêm feature", "milestone mới", "upgrade"
59
59
  - Treat all user text after the skill mention as `{{VP_ARGS}}`
@@ -62,8 +62,57 @@ Silent if command unavailable or errors.
62
62
  Prompt user conversationally with options.
63
63
 
64
64
  ## C. Tool Usage
65
- Use Cursor tools: `Shell`, `ReadFile`, `Glob`, `rg`, `ApplyPatch`, `WebSearch`, `WebFetch`, `Subagent`
66
- </cursor_skill_adapter>
65
+ Use Claude Code tools: `Bash` (shell), `Read` (file), `Edit` + `Write` (file write/patch),
66
+ `Grep` (search), `Glob` (file patterns), `LS`, `WebSearch`, `WebFetch`,
67
+ `Agent` (spawn subagent — multi-level nesting supported)
68
+ Interactive: `AskUserQuestion` (deferred — preload via ToolSearch before first call)
69
+ </adapter>
70
+
71
+ <adapter id="cursor-agent">
72
+ ## A. Skill Invocation
73
+ Same trigger keywords as claude-code adapter.
74
+
75
+ ## C. Tool Usage
76
+ Use Cursor tools: `run_terminal_cmd` (shell), `read_file` (read), `edit_file` (write/edit),
77
+ `grep_search` (search), `web_search`, `codebase_search`, `list_dir`, `file_search`
78
+ Interactive: text list fallback (AskQuestion available in Plan Mode only; Agent Mode = text)
79
+ Subagent: `/multitask` (user command, single-level only — not a callable tool)
80
+ MCP limit: 40 tools
81
+ </adapter>
82
+
83
+ <adapter id="antigravity">
84
+ ## A. Skill Invocation
85
+ Same trigger keywords as claude-code adapter.
86
+ Skill discovery: LLM-driven (automatic, no slash command needed).
87
+
88
+ ## C. Tool Usage
89
+ Use Antigravity tools: `shell` (cmd), `file_read`, `file_write`, MCP plugins
90
+ Interactive: text fallback (TUI-based; no formal AskUserQuestion)
91
+ Skill path: `.agents/skills/<skill>/SKILL.md` (project) or `~/.gemini/antigravity/skills/` (global)
92
+ Note: Gemini CLI deprecated June 18, 2026 — use Antigravity CLI.
93
+ </adapter>
94
+
95
+ <adapter id="codex">
96
+ ## A. Skill Invocation
97
+ Same trigger keywords as claude-code adapter.
98
+
99
+ ## C. Tool Usage
100
+ Use Codex tools: `container.exec` (sandboxed shell), `apply_patch` (file write), `web_search`
101
+ Interactive: text fallback (TUI Tab/Enter injection)
102
+ Config: `~/.codex/config.toml`
103
+ </adapter>
104
+
105
+ <adapter id="copilot">
106
+ ## A. Skill Invocation
107
+ Same trigger keywords as claude-code adapter.
108
+ Discovery: User-driven (`@agent-name` in GitHub Copilot Chat).
109
+
110
+ ## C. Tool Usage
111
+ Use Copilot tools: `runCommands` (shell), `read`/`readfile` (read), `edit`/`editFiles` (write),
112
+ `code_search`, `find_references`
113
+ Interactive: `askQuestions` (main agent only — NOT available in subagents; VS Code issue #293745)
114
+ Skill path: `.github/agents/<name>.agent.md`
115
+ </adapter>
67
116
  <scope_policy>
68
117
  ## ViePilot Namespace Guard (BUG-004)
69
118
  - Default mode: only use and reference `vp-*` skills in ViePilot workflows.
@@ -53,14 +53,63 @@ Silent if command unavailable or errors.
53
53
  </persona_context>
54
54
 
55
55
 
56
- <cursor_skill_adapter>
56
+ <adapter id="claude-code">
57
57
  ## A. Skill Invocation
58
58
  - Skill được gọi khi user mention `vp-info`, `/vp-info`, "viepilot version", "phiên bản viepilot", "skills list bundle"
59
59
  - Treat all user text after the skill mention as `{{VP_ARGS}}`
60
60
 
61
61
  ## B. Tool Usage
62
- Use Cursor tools: `Shell`, `ReadFile`, `Glob`, `rg`, `ApplyPatch`, `WebSearch`, `WebFetch`, `Subagent`
63
- </cursor_skill_adapter>
62
+ Use Claude Code tools: `Bash` (shell), `Read` (file), `Edit` + `Write` (file write/patch),
63
+ `Grep` (search), `Glob` (file patterns), `LS`, `WebSearch`, `WebFetch`,
64
+ `Agent` (spawn subagent — multi-level nesting supported)
65
+ Interactive: `AskUserQuestion` (deferred — preload via ToolSearch before first call)
66
+ </adapter>
67
+
68
+ <adapter id="cursor-agent">
69
+ ## A. Skill Invocation
70
+ Same trigger keywords as claude-code adapter.
71
+
72
+ ## C. Tool Usage
73
+ Use Cursor tools: `run_terminal_cmd` (shell), `read_file` (read), `edit_file` (write/edit),
74
+ `grep_search` (search), `web_search`, `codebase_search`, `list_dir`, `file_search`
75
+ Interactive: text list fallback (AskQuestion available in Plan Mode only; Agent Mode = text)
76
+ Subagent: `/multitask` (user command, single-level only — not a callable tool)
77
+ MCP limit: 40 tools
78
+ </adapter>
79
+
80
+ <adapter id="antigravity">
81
+ ## A. Skill Invocation
82
+ Same trigger keywords as claude-code adapter.
83
+ Skill discovery: LLM-driven (automatic, no slash command needed).
84
+
85
+ ## C. Tool Usage
86
+ Use Antigravity tools: `shell` (cmd), `file_read`, `file_write`, MCP plugins
87
+ Interactive: text fallback (TUI-based; no formal AskUserQuestion)
88
+ Skill path: `.agents/skills/<skill>/SKILL.md` (project) or `~/.gemini/antigravity/skills/` (global)
89
+ Note: Gemini CLI deprecated June 18, 2026 — use Antigravity CLI.
90
+ </adapter>
91
+
92
+ <adapter id="codex">
93
+ ## A. Skill Invocation
94
+ Same trigger keywords as claude-code adapter.
95
+
96
+ ## C. Tool Usage
97
+ Use Codex tools: `container.exec` (sandboxed shell), `apply_patch` (file write), `web_search`
98
+ Interactive: text fallback (TUI Tab/Enter injection)
99
+ Config: `~/.codex/config.toml`
100
+ </adapter>
101
+
102
+ <adapter id="copilot">
103
+ ## A. Skill Invocation
104
+ Same trigger keywords as claude-code adapter.
105
+ Discovery: User-driven (`@agent-name` in GitHub Copilot Chat).
106
+
107
+ ## C. Tool Usage
108
+ Use Copilot tools: `runCommands` (shell), `read`/`readfile` (read), `edit`/`editFiles` (write),
109
+ `code_search`, `find_references`
110
+ Interactive: `askQuestions` (main agent only — NOT available in subagents; VS Code issue #293745)
111
+ Skill path: `.github/agents/<name>.agent.md`
112
+ </adapter>
64
113
  <scope_policy>
65
114
  ## ViePilot Namespace Guard (BUG-004)
66
115
  - Default mode: only use and reference `vp-*` skills in ViePilot workflows.