agentpack-cli 0.3.13__tar.gz → 0.3.14__tar.gz

{agentpack_cli-0.3.13 → agentpack_cli-0.3.14}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agentpack-cli
-Version: 0.3.13
+Version: 0.3.14
 Summary: Local MCP context router for Claude Code, Codex, Cursor, and AI coding agents.
 License: MIT
 License-File: LICENSE
@@ -63,7 +63,7 @@ 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 (v0.3.14).** 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.
 >
@@ -205,10 +205,16 @@ 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`.
 
 ## Agent Setup
 
@@ -319,7 +325,7 @@ 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 |

{agentpack_cli-0.3.13 → agentpack_cli-0.3.14}/README.md

@@ -24,7 +24,7 @@ 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 (v0.3.14).** 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.
 >
@@ -166,10 +166,16 @@ 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`.
 
 ## Agent Setup
 
@@ -280,7 +286,7 @@ 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 |

{agentpack_cli-0.3.13 → agentpack_cli-0.3.14}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "agentpack-cli"
-version = "0.3.13"
+version = "0.3.14"
 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-0.3.14}/src/agentpack/__init__.py

@@ -1,3 +1,3 @@
 """AgentPack — task-aware context packing for AI coding agents."""
 
-__version__ = "0.3.13"
+__version__ = "0.3.14"

{agentpack_cli-0.3.13 → agentpack_cli-0.3.14}/src/agentpack/application/pack_service.py

@@ -660,6 +660,8 @@ 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,
             }
         )

{agentpack_cli-0.3.13 → agentpack_cli-0.3.14}/src/agentpack/commands/benchmark.py

@@ -469,6 +469,26 @@ def _run_git(cwd: Path | None, args: list[str]) -> None:
     )
 
 
+def _git_commit_exists(cwd: Path, commit: str) -> bool:
+    result = subprocess.run(
+        ["git", "cat-file", "-e", f"{commit}^{{commit}}"],
+        cwd=cwd,
+        text=True,
+        stdout=subprocess.PIPE,
+        stderr=subprocess.PIPE,
+        check=False,
+    )
+    return result.returncode == 0
+
+
+def _ensure_git_commit(cwd: Path, commit: str) -> None:
+    if _git_commit_exists(cwd, commit):
+        return
+    _run_git(cwd, ["fetch", "--quiet", "--depth", "1", "origin", commit])
+    if not _git_commit_exists(cwd, commit):
+        raise RuntimeError(f"Unable to fetch public benchmark commit {commit}")
+
+
 def _ensure_public_repo_clone(
     spec: PublicRepoSpec,
     cache_dir: Path,
@@ -512,7 +532,9 @@ def _run_public_repo_suite(
         for spec in specs:
             source_repo = _ensure_public_repo_clone(spec, cache, refresh=refresh)
             for public_case in spec.cases:
+                _ensure_git_commit(source_repo, public_case.commit)
                 parent = _git_stdout(source_repo, ["rev-parse", f"{public_case.commit}^"])
+                _ensure_git_commit(source_repo, parent)
                 work_root = temp_root / f"{spec.name}-{public_case.commit[:8]}"
                 shutil.copytree(
                     source_repo,

{agentpack_cli-0.3.13 → agentpack_cli-0.3.14}/src/agentpack/commands/dev_check.py

@@ -40,6 +40,8 @@ def register(app: typer.Typer) -> None:
                 marker = "[green]✓[/]" if item["returncode"] == 0 else "[red]✗[/]"
                 console.print(f"{marker} {item['name']} ({item['duration_s']:.2f}s)")
                 if item["returncode"] != 0:
+                    if item["output_excerpt"]:
+                        console.print(item["output_excerpt"])
                     console.print(f"  rerun: [bold]{item['command']}[/]")
         if failed:
             raise typer.Exit(1)
@@ -48,10 +50,21 @@ def register(app: typer.Typer) -> None:
 def _run(stage: CheckStage) -> dict[str, Any]:
     started = time.perf_counter()
     result = subprocess.run(stage.command, cwd=_root(), capture_output=True, text=True)
+    output = (result.stdout + "\n" + result.stderr).strip()
     return {
         "name": stage.name,
         "command": " ".join(stage.command),
         "returncode": result.returncode,
         "duration_s": round(time.perf_counter() - started, 3),
-        "detail": ((result.stderr or result.stdout).strip().splitlines() or [""])[-1],
+        "detail": (output.splitlines() or [""])[-1],
+        "output_excerpt": _output_excerpt(output) if result.returncode != 0 else "",
     }
+
+
+def _output_excerpt(output: str, *, max_lines: int = 80) -> str:
+    lines = output.splitlines()
+    if len(lines) <= max_lines:
+        excerpt = lines
+    else:
+        excerpt = ["... output truncated to final failing lines ...", *lines[-max_lines:]]
+    return "\n".join(f"  {line}" for line in excerpt)

{agentpack_cli-0.3.13 → agentpack_cli-0.3.14}/src/agentpack/commands/init.py

@@ -68,6 +68,8 @@ def _repo_gitignore_entries(share_cache: bool = False, agent: str = "generic") -
             ".agentpack/agent-lessons.md",
             ".agentpack/learning.prompt.md",
             ".agentpack/pr-learning-comment.md",
+            ".agentpack/learning-dashboard.html",
+            ".agentpack/team-lessons.md",
             ".agentpack/learning-feedback.jsonl",
             ".agentignore",
         ]
@@ -107,6 +109,8 @@ def _agentpack_gitignore_content(share_cache: bool = False) -> str:
             "agent-lessons.md",
             "learning.prompt.md",
             "pr-learning-comment.md",
+            "learning-dashboard.html",
+            "team-lessons.md",
             "learning-feedback.jsonl",
         ]
     )

agentpack_cli-0.3.14/src/agentpack/commands/learn.py

@@ -0,0 +1,179 @@
+from __future__ import annotations
+
+import json
+from datetime import datetime
+
+import typer
+
+from agentpack.commands._shared import _atomic_write, _root, console
+from agentpack.core.config import load_config
+from agentpack.learning.collector import collect_learning_inputs
+from agentpack.learning.extractor import build_learning_report
+from agentpack.learning.feedback import apply_feedback_to_report, load_feedback_summary, record_learning_feedback
+from agentpack.learning.lesson_ranker import rank_agent_lessons
+from agentpack.learning.provider import LearningProviderError, run_provider_command
+from agentpack.learning.quality import score_learning_report
+from agentpack.learning.renderers import (
+    learning_report_to_dict,
+    render_dashboard_html,
+    render_agent_lessons_markdown,
+    render_drills_markdown,
+    render_llm_prompt_markdown,
+    render_learning_markdown,
+    render_pr_comment_markdown,
+    render_provider_preview_markdown,
+    render_quality_markdown,
+    render_team_lessons_markdown,
+)
+from agentpack.learning.skill_map import apply_skill_feedback, recommend_practice_drills, render_skill_summary, update_skill_map
+
+
+def register(app: typer.Typer) -> None:
+    @app.command()
+    def learn(
+        task: str = typer.Option("auto", "--task", help="Task source. Only 'auto' is supported."),
+        since: str | None = typer.Option(None, "--since", help="Git ref to compare against, e.g. HEAD~1 or main."),
+        today: bool = typer.Option(False, "--today", help="Use today's work scope label for the report."),
+        output: str = typer.Option("", "--output", "-o", help="Markdown output path."),
+        json_output: bool = typer.Option(False, "--json", help="Print JSON to stdout instead of writing Markdown."),
+        llm_prompt: bool = typer.Option(False, "--llm-prompt", help="Write an LLM-ready learning prompt artifact."),
+        pr_comment: bool = typer.Option(False, "--pr-comment", help="Write a PR-comment-ready learning summary artifact."),
+        provider_preview: bool = typer.Option(False, "--provider-preview", help="Print the bounded provider payload without making a network call."),
+        provider_command: str = typer.Option("", "--provider-command", help="Run a local JSON-in/JSON-out provider command to enrich the report."),
+        dashboard: bool = typer.Option(False, "--dashboard", help="Write a static HTML learning dashboard artifact."),
+        team_export: bool = typer.Option(False, "--team-export", help="Write an opt-in team lesson export without personal skill history."),
+        ci: bool = typer.Option(False, "--ci", help="Fail when learning quality is below the configured threshold."),
+        skills: bool = typer.Option(False, "--skills", help="Print the local skill memory summary and exit."),
+        drills: bool = typer.Option(False, "--drills", help="Print recommended practice drills from local skill memory and exit."),
+        feedback: str = typer.Option("", "--feedback", help="Record feedback for this learning output (helpful|not-helpful)."),
+        feedback_note: str = typer.Option("", "--feedback-note", help="Optional note stored with --feedback."),
+        feedback_target: str = typer.Option("", "--feedback-target", help="Optional target such as skill:CLI design, lesson:retry, rename:old=>new, or merge:old=>new."),
+        suppress_skill: str = typer.Option("", "--suppress-skill", help="Suppress a noisy skill in future skill views and generation."),
+        rename_skill: str = typer.Option("", "--rename-skill", help="Rename a skill using old=>new."),
+        merge_skill: str = typer.Option("", "--merge-skill", help="Merge a skill using old=>new."),
+    ) -> None:
+        """Generate local learning artifacts from current task and git changes."""
+        if task != "auto":
+            console.print(
+                "[red]`agentpack learn --task \"...\"` is not supported. "
+                "Write .agentpack/task.md and use --task auto.[/]"
+            )
+            raise typer.Exit(2)
+
+        root = _root()
+        cfg = load_config(root)
+        skill_map_path = root / cfg.learning.skill_map_output
+        if skills:
+            typer.echo(render_skill_summary(skill_map_path), nl=False)
+            return
+        if drills:
+            typer.echo(render_drills_markdown(recommend_practice_drills(skill_map_path)), nl=False)
+            return
+        if suppress_skill:
+            apply_skill_feedback(skill_map_path, target=suppress_skill, action="suppress", note=feedback_note)
+            console.print(f"[green]✓[/] Suppressed skill {suppress_skill}")
+            return
+        if rename_skill:
+            old, new = _split_mapping(rename_skill, "--rename-skill")
+            apply_skill_feedback(skill_map_path, target=old, action="rename", replacement=new, note=feedback_note)
+            console.print(f"[green]✓[/] Renamed skill {old} -> {new}")
+            return
+        if merge_skill:
+            old, new = _split_mapping(merge_skill, "--merge-skill")
+            apply_skill_feedback(skill_map_path, target=old, action="merge", replacement=new, note=feedback_note)
+            console.print(f"[green]✓[/] Merged skill {old} -> {new}")
+            return
+
+        since_date = _today_start_iso() if today and not since else None
+        inputs = collect_learning_inputs(
+            root,
+            since=since,
+            since_date=since_date,
+            max_changed_files=cfg.learning.max_changed_files,
+            max_diff_chars_per_file=cfg.learning.max_diff_chars_per_file,
+        )
+        report = build_learning_report(
+            inputs,
+            max_cards=cfg.learning.max_cards,
+            max_quiz_questions=cfg.learning.max_quiz_questions,
+        )
+        feedback_summary = load_feedback_summary(root / cfg.learning.feedback_output)
+        report = apply_feedback_to_report(report, feedback_summary)
+        report.agent_lessons = rank_agent_lessons(report, feedback_summary, limit=cfg.learning.max_cards)
+        if today:
+            report.scope = "today"
+            if since_date:
+                report.since = f"today ({since_date})"
+
+        if provider_preview:
+            typer.echo(render_provider_preview_markdown(report), nl=False)
+            return
+
+        command = provider_command or cfg.learning.provider_command
+        if command:
+            try:
+                report = run_provider_command(command, report, timeout_seconds=cfg.learning.provider_timeout_seconds)
+            except LearningProviderError as exc:
+                console.print(f"[red]Provider command failed:[/] {exc}")
+                raise typer.Exit(1) from exc
+
+        update_skill_map(skill_map_path, report.skill_evidence)
+        agent_lessons_path = root / cfg.learning.agent_lessons_output
+        agent_lessons_path.parent.mkdir(parents=True, exist_ok=True)
+        _atomic_write(agent_lessons_path, render_agent_lessons_markdown(report))
+
+        quality = score_learning_report(report)
+        if quality.score < cfg.learning.min_groundedness_score:
+            console.print(f"[yellow]Learning quality warning:[/] score {quality.score}; " + "; ".join(quality.issues))
+
+        if llm_prompt:
+            prompt_path = root / cfg.learning.llm_prompt_output
+            prompt_path.parent.mkdir(parents=True, exist_ok=True)
+            _atomic_write(prompt_path, render_llm_prompt_markdown(report))
+        if pr_comment:
+            pr_path = root / cfg.learning.pr_comment_output
+            pr_path.parent.mkdir(parents=True, exist_ok=True)
+            _atomic_write(pr_path, render_pr_comment_markdown(report))
+        if dashboard:
+            dashboard_path = root / cfg.learning.dashboard_output
+            dashboard_path.parent.mkdir(parents=True, exist_ok=True)
+            _atomic_write(dashboard_path, render_dashboard_html(report))
+        if team_export:
+            team_path = root / cfg.learning.team_lessons_output
+            team_path.parent.mkdir(parents=True, exist_ok=True)
+            _atomic_write(team_path, render_team_lessons_markdown(report))
+        if feedback:
+            if feedback not in {"helpful", "not-helpful"}:
+                console.print("[red]--feedback must be helpful or not-helpful.[/]")
+                raise typer.Exit(2)
+            record_learning_feedback(root / cfg.learning.feedback_output, report, feedback, feedback_note, feedback_target)
+        if ci:
+            typer.echo(render_quality_markdown(report, quality.score, quality.issues), nl=False)
+            if quality.score < cfg.learning.min_groundedness_score:
+                raise typer.Exit(1)
+
+        if json_output:
+            typer.echo(json.dumps(learning_report_to_dict(report), indent=2, sort_keys=True))
+            return
+
+        default_output = cfg.learning.daily_output if today else cfg.learning.markdown_output
+        out_path = root / (output or default_output)
+        out_path.parent.mkdir(parents=True, exist_ok=True)
+        _atomic_write(out_path, render_learning_markdown(report))
+        console.print(f"[green]✓[/] Wrote {out_path.relative_to(root)}")
+
+
+def _today_start_iso() -> str:
+    now = datetime.now().astimezone()
+    return now.replace(hour=0, minute=0, second=0, microsecond=0).isoformat()
+
+
+def _split_mapping(value: str, flag: str) -> tuple[str, str]:
+    if "=>" not in value:
+        console.print(f"[red]{flag} expects old=>new.[/]")
+        raise typer.Exit(2)
+    old, new = [part.strip() for part in value.split("=>", 1)]
+    if not old or not new:
+        console.print(f"[red]{flag} expects non-empty old=>new values.[/]")
+        raise typer.Exit(2)
+    return old, new

{agentpack_cli-0.3.13 → agentpack_cli-0.3.14}/src/agentpack/commands/release_check.py

@@ -23,6 +23,7 @@ class StageResult:
     duration_s: float
     returncode: int = 0
     detail: str = ""
+    output_excerpt: str = ""
 
     def as_dict(self) -> dict[str, Any]:
         return {
@@ -32,6 +33,7 @@ class StageResult:
             "duration_s": round(self.duration_s, 3),
             "returncode": self.returncode,
             "detail": self.detail,
+            "output_excerpt": self.output_excerpt,
         }
 
 
@@ -64,6 +66,8 @@ def register(app: typer.Typer) -> None:
                 console.print(f"{marker} {stage.name}: {stage.status} ({stage.duration_s:.2f}s)")
                 if stage.detail and stage.status != "passed":
                     console.print(f"  {stage.detail}")
+                if stage.output_excerpt and stage.status != "passed":
+                    console.print(stage.output_excerpt)
                 if stage.status != "passed":
                     console.print(f"  rerun: [bold]{stage.command}[/]")
         if failed:
@@ -93,7 +97,8 @@ def _run_stage(root: Path, name: str, command: list[str]) -> StageResult:
         result = subprocess.run(command, cwd=root, capture_output=True, text=True)
     except OSError as exc:
         return StageResult(name=name, command=" ".join(command), status="failed", duration_s=time.perf_counter() - started, returncode=1, detail=str(exc))
-    output = (result.stderr or result.stdout).strip().splitlines()
+    combined_output = (result.stdout + "\n" + result.stderr).strip()
+    output = combined_output.splitlines()
     return StageResult(
         name=name,
         command=" ".join(command),
@@ -101,4 +106,14 @@ def _run_stage(root: Path, name: str, command: list[str]) -> StageResult:
         duration_s=time.perf_counter() - started,
         returncode=result.returncode,
         detail=output[-1] if output else "",
+        output_excerpt=_output_excerpt(combined_output) if result.returncode != 0 else "",
     )
+
+
+def _output_excerpt(output: str, *, max_lines: int = 80) -> str:
+    lines = output.splitlines()
+    if len(lines) <= max_lines:
+        excerpt = lines
+    else:
+        excerpt = ["... output truncated to final failing lines ...", *lines[-max_lines:]]
+    return "\n".join(f"  {line}" for line in excerpt)

{agentpack_cli-0.3.13 → agentpack_cli-0.3.14}/src/agentpack/core/config.py

@@ -55,6 +55,10 @@ class LearningConfig(BaseModel):
     llm_prompt_output: str = ".agentpack/learning.prompt.md"
     pr_comment_output: str = ".agentpack/pr-learning-comment.md"
     feedback_output: str = ".agentpack/learning-feedback.jsonl"
+    dashboard_output: str = ".agentpack/learning-dashboard.html"
+    team_lessons_output: str = ".agentpack/team-lessons.md"
+    provider_command: str = ""
+    provider_timeout_seconds: int = 60
     inject_agent_lessons: bool = True
     max_changed_files: int = 20
     max_diff_chars_per_file: int = 1200
@@ -184,6 +188,10 @@ agent_lessons_output = ".agentpack/agent-lessons.md"
 llm_prompt_output = ".agentpack/learning.prompt.md"
 pr_comment_output = ".agentpack/pr-learning-comment.md"
 feedback_output = ".agentpack/learning-feedback.jsonl"
+dashboard_output = ".agentpack/learning-dashboard.html"
+team_lessons_output = ".agentpack/team-lessons.md"
+provider_command = ""
+provider_timeout_seconds = 60
 inject_agent_lessons = true
 max_changed_files = 20
 max_diff_chars_per_file = 1200

{agentpack_cli-0.3.13 → agentpack_cli-0.3.14}/src/agentpack/learning/__init__.py

@@ -1,5 +1,7 @@
 from agentpack.learning.models import (
     AgentLesson,
+    FeedbackSignal,
+    FeedbackSummary,
     LearningCard,
     LearningOptions,
     LearningReport,
@@ -11,6 +13,8 @@ from agentpack.learning.models import (
 
 __all__ = [
     "AgentLesson",
+    "FeedbackSignal",
+    "FeedbackSummary",
     "LearningCard",
     "LearningOptions",
     "LearningReport",

agentpack_cli-0.3.14/src/agentpack/learning/feedback.py

@@ -0,0 +1,171 @@
+from __future__ import annotations
+
+import json
+from datetime import datetime, timezone
+from pathlib import Path
+
+from pydantic import ValidationError
+
+from agentpack.learning.models import FeedbackSignal, FeedbackSummary, LearningReport
+
+
+def record_learning_feedback(
+    path: Path,
+    report: LearningReport,
+    feedback: str,
+    note: str = "",
+    target: str = "",
+) -> None:
+    path.parent.mkdir(parents=True, exist_ok=True)
+    payload = FeedbackSignal(
+        created_at=datetime.now(timezone.utc).isoformat(),
+        task=report.task,
+        scope=report.scope,
+        feedback=feedback,
+        note=note,
+        target=target,
+        concepts=report.concepts,
+        source_files=[source.path for source in report.source_files],
+    ).model_dump(mode="json")
+    with path.open("a", encoding="utf-8") as fh:
+        fh.write(json.dumps(payload, sort_keys=True) + "\n")
+
+
+def load_feedback_summary(path: Path) -> FeedbackSummary:
+    summary = FeedbackSummary()
+    if not path.exists():
+        return summary
+    for line in path.read_text(encoding="utf-8").splitlines():
+        if not line.strip():
+            continue
+        try:
+            signal = FeedbackSignal.model_validate_json(line)
+        except (ValidationError, ValueError):
+            continue
+        _apply_signal(summary, signal)
+    return summary
+
+
+def apply_feedback_to_report(report: LearningReport, summary: FeedbackSummary) -> LearningReport:
+    if not _has_feedback(summary):
+        return report
+
+    rename_map = {old.lower(): new for old, new in summary.skill_renames.items()}
+    merge_map = {old.lower(): new for old, new in summary.skill_merges.items()}
+    suppressed_skills = {skill.lower() for skill in summary.suppressed_skills}
+    not_helpful = {concept.lower() for concept in summary.not_helpful_concepts}
+    suppressed = suppressed_skills | not_helpful
+
+    for source in report.source_files:
+        source.concepts = [_normalize_concept(concept, rename_map, merge_map) for concept in source.concepts]
+        source.concepts = [concept for concept in source.concepts if concept.lower() not in suppressed]
+
+    report.concepts = [_normalize_concept(concept, rename_map, merge_map) for concept in report.concepts]
+    report.concepts = _dedupe([concept for concept in report.concepts if concept.lower() not in suppressed])
+
+    report.learning_cards = [
+        card
+        for card in report.learning_cards
+        if card.title.lower() not in suppressed
+        and not any(concept.lower() == card.title.lower() for concept in summary.not_helpful_concepts)
+    ]
+    for card in report.learning_cards:
+        normalized = _normalize_concept(card.title, rename_map, merge_map)
+        if normalized != card.title:
+            card.title = normalized
+        if card.title.lower() in {item.lower() for item in summary.helpful_concepts}:
+            card.body = f"{card.body} Prior feedback marked this concept useful; keep practicing it with changed-file evidence."
+
+    report.agent_lessons = [
+        lesson
+        for lesson in report.agent_lessons
+        if not _lesson_suppressed(lesson.rule, summary)
+    ]
+    for lesson in report.agent_lessons:
+        if _lesson_helpful(lesson.rule, report.concepts, summary):
+            lesson.status = "accepted"
+            lesson.confidence = min(100, max(lesson.confidence, 90))
+
+    for item in report.skill_evidence:
+        item.skill = _normalize_concept(item.skill, rename_map, merge_map)
+        if item.skill.lower() in {concept.lower() for concept in summary.helpful_concepts}:
+            item.confidence = min(100, item.confidence + 10)
+    report.skill_evidence = [
+        item for item in report.skill_evidence if item.skill.lower() not in suppressed
+    ]
+    return report
+
+
+def _apply_signal(summary: FeedbackSummary, signal: FeedbackSignal) -> None:
+    target = signal.target.strip()
+    target_kind, _, target_value = target.partition(":")
+    normalized_feedback = signal.feedback.strip().lower()
+    if normalized_feedback == "helpful":
+        summary.helpful_concepts.update(signal.concepts)
+        if signal.note:
+            summary.accepted_notes.append(signal.note)
+    elif normalized_feedback == "not-helpful":
+        summary.not_helpful_concepts.update(signal.concepts)
+
+    if target_kind == "skill" and target_value:
+        if normalized_feedback == "not-helpful":
+            summary.suppressed_skills.add(target_value)
+        elif normalized_feedback == "helpful":
+            summary.helpful_concepts.add(target_value)
+    elif target_kind == "lesson" and target_value and normalized_feedback == "not-helpful":
+        summary.suppressed_lesson_terms.add(target_value)
+    elif target_kind == "rename" and "=>" in target_value:
+        old, new = [part.strip() for part in target_value.split("=>", 1)]
+        if old and new:
+            summary.skill_renames[old] = new
+    elif target_kind == "merge" and "=>" in target_value:
+        old, new = [part.strip() for part in target_value.split("=>", 1)]
+        if old and new:
+            summary.skill_merges[old] = new
+
+
+def _has_feedback(summary: FeedbackSummary) -> bool:
+    return bool(
+        summary.helpful_concepts
+        or summary.not_helpful_concepts
+        or summary.suppressed_skills
+        or summary.suppressed_lesson_terms
+        or summary.skill_renames
+        or summary.skill_merges
+    )
+
+
+def _normalize_concept(concept: str, renames: dict[str, str], merges: dict[str, str]) -> str:
+    key = concept.lower()
+    return merges.get(key) or renames.get(key) or concept
+
+
+def _lesson_suppressed(rule: str, summary: FeedbackSummary) -> bool:
+    lower = rule.lower()
+    return any(term.lower() in lower for term in summary.suppressed_lesson_terms)
+
+
+def _lesson_helpful(rule: str, concepts: list[str], summary: FeedbackSummary) -> bool:
+    lower_rule = rule.lower()
+    helpful = {concept.lower() for concept in summary.helpful_concepts}
+    for concept in concepts:
+        lower_concept = concept.lower()
+        if lower_concept not in helpful:
+            continue
+        if lower_concept in lower_rule:
+            return True
+        if any(part and part in lower_rule for part in lower_concept.split()):
+            return True
+    return False
+
+
+def _dedupe(values: list[str]) -> list[str]:
+    seen: set[str] = set()
+    result: list[str] = []
+    for value in values:
+        key = value.lower()
+        if key in seen:
+            continue
+        seen.add(key)
+        result.append(value)
+    return result