agentpack-cli 0.3.13__tar.gz → 0.3.15__tar.gz

{agentpack_cli-0.3.13 → agentpack_cli-0.3.15}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agentpack-cli
-Version: 0.3.13
+Version: 0.3.15
 Summary: Local MCP context router for Claude Code, Codex, Cursor, and AI coding agents.
 License: MIT
 License-File: LICENSE
@@ -63,12 +63,39 @@ pipx run --spec agentpack-cli agentpack route --task "fix auth token expiry"
 
 ![AgentPack route demo](docs/assets/agentpack-route-demo.svg)
 
-> **Status: alpha (v0.3.13).** Works, tested, and used in real sessions. Python and JavaScript/TypeScript are the best-supported languages. Current benchmarks are useful regression checks, not broad proof that AgentPack improves coding-agent success. API may change before 1.0.
+> **Status: alpha (v0.3.15).** Works, tested, and used in real sessions. Python and JavaScript/TypeScript are the best-supported languages. Current benchmarks are useful regression checks, not broad proof that AgentPack improves coding-agent success. API may change before 1.0.
 >
 > **Platform note:** macOS, Linux, and Windows are supported. Windows support targets PowerShell plus Git for Windows. `cmd.exe` and bare Git setups are not a supported path yet.
 >
 > **Name note:** PyPI package is `agentpack-cli`, npm package is `@vishal2612200/agentpack`, and the command is `agentpack`. This project is unrelated to AgentPack dataset papers or other repos with the same name.
 
+## What's New in 0.3.15
+
+AgentPack Router now recommends skills with stronger local signals: richer
+frontmatter, confidence thresholds, negative triggers, diversity-aware ranking,
+and a pull-based MCP `get_skill` flow. Skill benchmark cases can declare
+`expected_skills` and `avoid_skills`, and `agentpack skills feedback` records
+local outcome signals for future routing.
+
+## What's New in 0.3.14
+
+AgentPack Learn now covers both sides of AI-assisted development: the coding
+agent gets compact future-agent lessons, and the developer gets task-specific
+learning notes, skill evidence, and practice follow-up.
+
+- `agentpack learn --provider-command` adds an opt-in local provider bridge:
+  AgentPack sends a bounded, redacted report JSON on stdin and accepts
+  LearningReport-compatible JSON fields on stdout. No hosted service is called
+  unless your command does it.
+- `agentpack learn --dashboard` writes a static
+  `.agentpack/learning-dashboard.html` for IDE/browser review.
+- `agentpack learn --team-export` writes `.agentpack/team-lessons.md`, a
+  shareable lesson file that omits personal skill history.
+- `agentpack learn --feedback`, `--skills`, and `--drills` close the loop from
+  a task summary to skill memory and next-practice prompts.
+- `agentpack dev-check` and `agentpack release-check` now print bounded failure
+  excerpts, so CI shows the failing test instead of only a red stage name.
+
 ## Before vs After
 
 Without AgentPack, a cold coding-agent session often starts with manual repo orientation:
@@ -205,10 +232,21 @@ agentpack learn --today
 agentpack learn --since main
 agentpack learn --json
 agentpack learn --llm-prompt --pr-comment
-agentpack learn --feedback helpful --feedback-note "Useful review prompts"
+agentpack learn --provider-preview
+agentpack learn --provider-command "python scripts/learn_provider.py"
+agentpack learn --dashboard --team-export
+agentpack learn --skills
+agentpack learn --drills
+agentpack learn --ci
+agentpack learn --feedback helpful --feedback-target "skill:CLI design" --feedback-note "Useful review prompts"
 ```
 
-AgentPack writes developer notes to `.agentpack/learning.md` or `.agentpack/daily-summary.md`, updates `.agentpack/skills-progress.json`, writes `.agentpack/agent-lessons.md` for future coding agents, and can emit `.agentpack/learning.prompt.md` or `.agentpack/pr-learning-comment.md`. The MVP is local-first and reuses AgentPack redaction before reading diff snippets.
+AgentPack writes developer notes to `.agentpack/learning.md` or `.agentpack/daily-summary.md`, updates a local skill memory in `.agentpack/skills-progress.json`, writes ranked `.agentpack/agent-lessons.md` for future coding agents, and can emit `.agentpack/learning.prompt.md`, `.agentpack/pr-learning-comment.md`, `.agentpack/learning-dashboard.html`, or `.agentpack/team-lessons.md`. Learn is local-first by default: `--provider-preview` shows the bounded payload for optional external refinement without making a network call, `--provider-command` runs only the local command you provide, and feedback stays in `.agentpack/learning-feedback.jsonl`.
+
+Use `--dashboard` when a developer wants an IDE-friendly review surface. Use
+`--team-export` when the useful lesson should be shared without publishing a
+personal skill history. Use `--ci` to fail a workflow when the generated
+learning is too generic or lacks changed-file evidence.
 
 ## Agent Setup
 
@@ -247,10 +285,15 @@ Generated instructions keep thread mode explicit. They recommend `AGENTPACK_THRE
 default_mode = "balanced"
 default_budget = 40000
 
+[context_lite]
+budget = 8000
+
 [agents.generic]
 output = ".agentpack/context.md"
 ```
 
+Use `agentpack pack --mode lite` when you want a cheap ranked map before deeper file reads. Use `minimal`, `balanced`, or `deep` when you want progressively more file content in the generated pack.
+
 Use `.agentignore` to remove generated output, vendored code, large exports, or files that repeatedly appear as ranking noise. AgentPack imports obvious generated/noisy entries from gitignore sources during init, but repository-specific outputs should still be added by hand.
 
 Use scoring weights only after measuring a real miss:
@@ -319,7 +362,7 @@ gate.
 | `agentpack work "task"` | Initialize if needed, start task, refresh context, show next steps |
 | `agentpack start "task"` | Write task and run the guard/refresh workflow |
 | `agentpack finish --since main` | Diagnose, capture benchmark case, run checks, mark done |
-| `agentpack learn` | Generate developer learning notes, skill progress, and future-agent lessons |
+| `agentpack learn` | Generate developer learning notes, skill memory, feedback-aware drills, and future-agent lessons |
 | `agentpack task show|set|clear` | Manage global or thread-scoped task files |
 | `agentpack pack` | Generate a ranked context pack for `.agentpack/task.md` |
 | `agentpack next --fix-all-safe` | Ask AgentPack what command or safe repair should happen next |
@@ -408,6 +451,8 @@ AgentPack writes local artifacts under `.agentpack/`:
 | `.agentpack/agent-lessons.md` | compact repo-specific lessons injected into future packs |
 | `.agentpack/learning.prompt.md` | optional source-backed prompt for external LLM refinement |
 | `.agentpack/pr-learning-comment.md` | optional PR-comment-ready learning summary |
+| `.agentpack/learning-dashboard.html` | optional static dashboard from `agentpack learn --dashboard` |
+| `.agentpack/team-lessons.md` | optional shared lesson export from `agentpack learn --team-export` |
 | `.agentpack/learning-feedback.jsonl` | optional local helpful/not-helpful feedback records |
 | `.agentpack/pack_metadata.json` | freshness and pack metadata |
 | `.agentpack/cache/` | offline file summaries keyed by hash |

{agentpack_cli-0.3.13 → agentpack_cli-0.3.15}/README.md

@@ -24,12 +24,39 @@ pipx run --spec agentpack-cli agentpack route --task "fix auth token expiry"
 
 ![AgentPack route demo](docs/assets/agentpack-route-demo.svg)
 
-> **Status: alpha (v0.3.13).** Works, tested, and used in real sessions. Python and JavaScript/TypeScript are the best-supported languages. Current benchmarks are useful regression checks, not broad proof that AgentPack improves coding-agent success. API may change before 1.0.
+> **Status: alpha (v0.3.15).** Works, tested, and used in real sessions. Python and JavaScript/TypeScript are the best-supported languages. Current benchmarks are useful regression checks, not broad proof that AgentPack improves coding-agent success. API may change before 1.0.
 >
 > **Platform note:** macOS, Linux, and Windows are supported. Windows support targets PowerShell plus Git for Windows. `cmd.exe` and bare Git setups are not a supported path yet.
 >
 > **Name note:** PyPI package is `agentpack-cli`, npm package is `@vishal2612200/agentpack`, and the command is `agentpack`. This project is unrelated to AgentPack dataset papers or other repos with the same name.
 
+## What's New in 0.3.15
+
+AgentPack Router now recommends skills with stronger local signals: richer
+frontmatter, confidence thresholds, negative triggers, diversity-aware ranking,
+and a pull-based MCP `get_skill` flow. Skill benchmark cases can declare
+`expected_skills` and `avoid_skills`, and `agentpack skills feedback` records
+local outcome signals for future routing.
+
+## What's New in 0.3.14
+
+AgentPack Learn now covers both sides of AI-assisted development: the coding
+agent gets compact future-agent lessons, and the developer gets task-specific
+learning notes, skill evidence, and practice follow-up.
+
+- `agentpack learn --provider-command` adds an opt-in local provider bridge:
+  AgentPack sends a bounded, redacted report JSON on stdin and accepts
+  LearningReport-compatible JSON fields on stdout. No hosted service is called
+  unless your command does it.
+- `agentpack learn --dashboard` writes a static
+  `.agentpack/learning-dashboard.html` for IDE/browser review.
+- `agentpack learn --team-export` writes `.agentpack/team-lessons.md`, a
+  shareable lesson file that omits personal skill history.
+- `agentpack learn --feedback`, `--skills`, and `--drills` close the loop from
+  a task summary to skill memory and next-practice prompts.
+- `agentpack dev-check` and `agentpack release-check` now print bounded failure
+  excerpts, so CI shows the failing test instead of only a red stage name.
+
 ## Before vs After
 
 Without AgentPack, a cold coding-agent session often starts with manual repo orientation:
@@ -166,10 +193,21 @@ agentpack learn --today
 agentpack learn --since main
 agentpack learn --json
 agentpack learn --llm-prompt --pr-comment
-agentpack learn --feedback helpful --feedback-note "Useful review prompts"
+agentpack learn --provider-preview
+agentpack learn --provider-command "python scripts/learn_provider.py"
+agentpack learn --dashboard --team-export
+agentpack learn --skills
+agentpack learn --drills
+agentpack learn --ci
+agentpack learn --feedback helpful --feedback-target "skill:CLI design" --feedback-note "Useful review prompts"
 ```
 
-AgentPack writes developer notes to `.agentpack/learning.md` or `.agentpack/daily-summary.md`, updates `.agentpack/skills-progress.json`, writes `.agentpack/agent-lessons.md` for future coding agents, and can emit `.agentpack/learning.prompt.md` or `.agentpack/pr-learning-comment.md`. The MVP is local-first and reuses AgentPack redaction before reading diff snippets.
+AgentPack writes developer notes to `.agentpack/learning.md` or `.agentpack/daily-summary.md`, updates a local skill memory in `.agentpack/skills-progress.json`, writes ranked `.agentpack/agent-lessons.md` for future coding agents, and can emit `.agentpack/learning.prompt.md`, `.agentpack/pr-learning-comment.md`, `.agentpack/learning-dashboard.html`, or `.agentpack/team-lessons.md`. Learn is local-first by default: `--provider-preview` shows the bounded payload for optional external refinement without making a network call, `--provider-command` runs only the local command you provide, and feedback stays in `.agentpack/learning-feedback.jsonl`.
+
+Use `--dashboard` when a developer wants an IDE-friendly review surface. Use
+`--team-export` when the useful lesson should be shared without publishing a
+personal skill history. Use `--ci` to fail a workflow when the generated
+learning is too generic or lacks changed-file evidence.
 
 ## Agent Setup
 
@@ -208,10 +246,15 @@ Generated instructions keep thread mode explicit. They recommend `AGENTPACK_THRE
 default_mode = "balanced"
 default_budget = 40000
 
+[context_lite]
+budget = 8000
+
 [agents.generic]
 output = ".agentpack/context.md"
 ```
 
+Use `agentpack pack --mode lite` when you want a cheap ranked map before deeper file reads. Use `minimal`, `balanced`, or `deep` when you want progressively more file content in the generated pack.
+
 Use `.agentignore` to remove generated output, vendored code, large exports, or files that repeatedly appear as ranking noise. AgentPack imports obvious generated/noisy entries from gitignore sources during init, but repository-specific outputs should still be added by hand.
 
 Use scoring weights only after measuring a real miss:
@@ -280,7 +323,7 @@ gate.
 | `agentpack work "task"` | Initialize if needed, start task, refresh context, show next steps |
 | `agentpack start "task"` | Write task and run the guard/refresh workflow |
 | `agentpack finish --since main` | Diagnose, capture benchmark case, run checks, mark done |
-| `agentpack learn` | Generate developer learning notes, skill progress, and future-agent lessons |
+| `agentpack learn` | Generate developer learning notes, skill memory, feedback-aware drills, and future-agent lessons |
 | `agentpack task show|set|clear` | Manage global or thread-scoped task files |
 | `agentpack pack` | Generate a ranked context pack for `.agentpack/task.md` |
 | `agentpack next --fix-all-safe` | Ask AgentPack what command or safe repair should happen next |
@@ -369,6 +412,8 @@ AgentPack writes local artifacts under `.agentpack/`:
 | `.agentpack/agent-lessons.md` | compact repo-specific lessons injected into future packs |
 | `.agentpack/learning.prompt.md` | optional source-backed prompt for external LLM refinement |
 | `.agentpack/pr-learning-comment.md` | optional PR-comment-ready learning summary |
+| `.agentpack/learning-dashboard.html` | optional static dashboard from `agentpack learn --dashboard` |
+| `.agentpack/team-lessons.md` | optional shared lesson export from `agentpack learn --team-export` |
 | `.agentpack/learning-feedback.jsonl` | optional local helpful/not-helpful feedback records |
 | `.agentpack/pack_metadata.json` | freshness and pack metadata |
 | `.agentpack/cache/` | offline file summaries keyed by hash |

{agentpack_cli-0.3.13 → agentpack_cli-0.3.15}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "agentpack-cli"
-version = "0.3.13"
+version = "0.3.15"
 description = "Local MCP context router for Claude Code, Codex, Cursor, and AI coding agents."
 readme = "README.md"
 requires-python = ">=3.10"

{agentpack_cli-0.3.13 → agentpack_cli-0.3.15}/src/agentpack/__init__.py

@@ -1,3 +1,3 @@
 """AgentPack — task-aware context packing for AI coding agents."""
 
-__version__ = "0.3.13"
+__version__ = "0.3.15"

{agentpack_cli-0.3.13 → agentpack_cli-0.3.15}/src/agentpack/application/pack_service.py

@@ -294,7 +294,6 @@ def _scan_metadata(
         "ignore_hash": _hash_text(ignore_text),
         "include_globs": include_globs,
         "exclude_globs": exclude_globs,
-        "generated_paths": sorted(generated_paths),
         "max_file_tokens": cfg.context.max_file_tokens,
         "workspace": workspace,
     }
@@ -329,7 +328,10 @@ def _full_scan_reason(
         return "git unavailable"
 
     previous_meta = previous_snap.get("metadata") if isinstance(previous_snap.get("metadata"), dict) else {}
-    if previous_meta.get("scan_fingerprint") != scan_metadata.get("scan_fingerprint"):
+    if previous_meta.get("scan_fingerprint") != scan_metadata.get("scan_fingerprint") and (
+        _normalized_scan_config(previous_meta.get("scan_config"))
+        != _normalized_scan_config(scan_metadata.get("scan_config"))
+    ):
         return "scan config or ignore rules changed"
     if previous_meta.get("git_branch") != scan_metadata.get("git_branch"):
         return "git branch changed"
@@ -346,6 +348,14 @@ def _hash_text(value: str) -> str:
     return hashlib.sha256(value.encode("utf-8")).hexdigest()
 
 
+def _normalized_scan_config(value: Any) -> dict[str, Any]:
+    if not isinstance(value, dict):
+        return {}
+    normalized = dict(value)
+    normalized.pop("generated_paths", None)
+    return normalized
+
+
 def _hash_json(value: dict[str, Any]) -> str:
     raw = json.dumps(value, sort_keys=True, separators=(",", ":"))
     return _hash_text(raw)
@@ -382,7 +392,7 @@ class PackPlanner:
     def plan(self, request: PackRequest) -> PackPlan:
         root = request.root
         cfg = load_config(root)
-        effective_budget = request.budget if request.budget > 0 else cfg.context.default_budget
+        effective_budget = _resolve_effective_budget(request, cfg)
         ignore_spec = load_spec(root / cfg.project.ignore_file)
         phase_times: dict[str, float] = {}
 
@@ -660,6 +670,8 @@ class AdapterRegistry:
                 cfg.learning.agent_lessons_output,
                 cfg.learning.llm_prompt_output,
                 cfg.learning.pr_comment_output,
+                cfg.learning.dashboard_output,
+                cfg.learning.team_lessons_output,
                 cfg.learning.feedback_output,
             }
         )
@@ -972,10 +984,18 @@ def _rank_omitted_relevant_files(files: list[OmittedRelevantFile]) -> list[Omitt
 
 
 def _repo_map_budget_for_mode(mode: str, effective_budget: int) -> int:
-    caps = {"minimal": 300, "balanced": 600, "deep": 900}
+    caps = {"lite": 150, "minimal": 300, "balanced": 600, "deep": 900}
     return min(caps.get(mode, 500), max(0, effective_budget // 20))
 
 
+def _resolve_effective_budget(request: PackRequest, cfg: Any) -> int:
+    if request.budget > 0:
+        return request.budget
+    if request.mode == "lite":
+        return cfg.context_lite.budget
+    return cfg.context.default_budget
+
+
 _FRONTEND_TASK_TERMS = {
     "component", "components", "frontend", "landing", "layout", "page", "pages",
     "preview", "previews", "public", "seo", "signup", "tool", "tools", "ui", "web",
@@ -1269,7 +1289,10 @@ def _summary_score_floor(cfg: Any, generic_ratio: float) -> float:
 
 
 def _summary_cap_for_mode(cfg: Any, mode: str, generic_ratio: float = 0.0) -> int:
-    if mode == "minimal":
+    if mode == "lite":
+        minimal_cap = cfg.context.max_summary_files_minimal
+        cap = min(minimal_cap, cfg.context_lite.max_selected_files) if minimal_cap > 0 else cfg.context_lite.max_selected_files
+    elif mode == "minimal":
         cap = cfg.context.max_summary_files_minimal
     elif mode == "balanced":
         cap = cfg.context.max_summary_files_balanced
@@ -1321,17 +1344,17 @@ def _guarded_summary_cap(
     if rows < 3:
         if no_live_changes and cap > 0:
             if effective_budget and effective_budget <= 2500:
-                return min(cap, 4 if mode == "minimal" else 6)
+                return min(cap, 4 if mode in ("lite", "minimal") else 6)
             if effective_budget and effective_budget <= 6000:
-                return min(cap, 12 if mode == "minimal" else 16)
+                return min(cap, 12 if mode in ("lite", "minimal") else 16)
             return min(cap, 16)
         return cap
     if avg_summary_precision <= 0.05:
         if no_live_changes:
             return -1
-        strict_cap = 3 if mode == "minimal" else 5 if mode == "balanced" else 10
+        strict_cap = 2 if mode == "lite" else 3 if mode == "minimal" else 5 if mode == "balanced" else 10
     elif avg_summary_precision <= 0.15:
-        strict_cap = 3 if no_live_changes else 6 if mode == "minimal" else 12 if mode == "balanced" else 20
+        strict_cap = 2 if mode == "lite" else 3 if no_live_changes else 6 if mode == "minimal" else 12 if mode == "balanced" else 20
     else:
         if no_live_changes and cap > 0:
             return min(cap, 8)
@@ -1389,15 +1412,15 @@ def _guarded_weak_signal_cap(
     if not no_live_changes:
         return 0
     if generic_ratio >= 0.5:
-        base = {"minimal": 0, "balanced": 1, "deep": 2}.get(mode, 1)
+        base = {"lite": 0, "minimal": 0, "balanced": 1, "deep": 2}.get(mode, 1)
     elif generic_ratio >= 0.35:
-        base = {"minimal": 1, "balanced": 2, "deep": 3}.get(mode, 2)
+        base = {"lite": 0, "minimal": 1, "balanced": 2, "deep": 3}.get(mode, 2)
     else:
-        base = {"minimal": 2, "balanced": 4, "deep": 6}.get(mode, 3)
+        base = {"lite": 1, "minimal": 2, "balanced": 4, "deep": 6}.get(mode, 3)
     avg_precision, rows = _recent_token_precision(root)
     if rows >= 3:
         if avg_precision <= 0.1:
-            base = min(base, 0 if mode == "minimal" else 1)
+            base = min(base, 0 if mode in ("lite", "minimal") else 1)
         elif avg_precision <= 0.2:
             base = min(base, 1 if mode != "deep" else 2)
     if effective_budget and effective_budget <= 2500: