loki-mode 7.12.0 → 7.14.0

package/SKILL.md

@@ -3,7 +3,7 @@ name: loki-mode
 description: Autonomous spec-to-product system. Triggers on "Loki Mode". Takes a spec (PRD, GitHub issue, OpenAPI doc, etc.) to deployed product via the RARV-C closure loop, with minimal human intervention. Provider-agnostic. Requires --dangerously-skip-permissions flag.
 ---
 
-# Loki Mode v7.12.0
+# Loki Mode v7.14.0
 
 **You are an autonomous agent. You make decisions. You do not ask questions. You do not stop.**
 
@@ -160,6 +160,8 @@ GROWTH ──[continuous improvement loop]──> GROWTH
 | `.loki/signals/HUMAN_REVIEW_NEEDED` | Never | When human decision required |
 | `.loki/state/checkpoints/` | After task completion | Automatic + manual via `loki checkpoint` |
 
+One-command rollback (v7.5.2+): `loki rollback latest` or `loki rollback to <id>` restores `.loki/` state from a checkpoint. It first captures a forced pre-rollback snapshot of the current state and prints its id, so a rollback is itself undoable (`loki rollback to <that-id>`). Use `loki rollback list` to see checkpoints.
+
 ---
 
 ## Module Loading Protocol (Skills)
@@ -381,4 +383,4 @@ See `CHANGELOG.md` entries [7.5.7], [7.5.8], [7.5.13] for the per-fix list and r
 
 ---
 
-**v7.12.0 | [Autonomi](https://www.autonomi.dev/) flagship product | ~260 lines core**
+**v7.14.0 | [Autonomi](https://www.autonomi.dev/) flagship product | ~260 lines core**

package/VERSION

@@ -1 +1 @@
-7.12.0
+7.14.0

package/autonomy/lib/wiki-ask.py

@@ -0,0 +1,137 @@
+#!/usr/bin/env python3
+"""wiki-ask.py -- grounded, cited codebase Q&A for the R5 auto-wiki.
+
+Answers a natural-language question about the project, grounded in the indexed
+codebase. Citations are file:line and ALWAYS point at real code:
+
+  1. Deterministically retrieve the top-K relevant chunks (token overlap).
+  2. Show the LLM NUMBERED chunks and tell it to cite by index only ([1]..[K]).
+  3. Map each [n] back to the real {file,start_line} of the chunk we supplied.
+  4. Validate every citation against the filesystem; drop non-resolving ones.
+
+A fabricated citation is structurally impossible: the model can only reference
+chunks we handed it, and only those that resolve on disk survive.
+
+LLM is mocked via LOKI_WIKI_LLM_STUB (see wiki_llm.py) so CI makes no paid calls.
+With no provider and no stub, an EXTRACTIVE answer is returned (top chunk
+snippets with their real citations).
+
+Invoked as a subprocess (hyphen in filename):
+  python3 wiki-ask.py --root <project> --question "..." [--k N] [--json] [--quiet]
+Exit codes: 0 answered, 2 usage/error, 3 no relevant code found.
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import os
+import sys
+from pathlib import Path
+
+_HERE = os.path.dirname(os.path.abspath(__file__))
+if _HERE not in sys.path:
+    sys.path.insert(0, _HERE)
+
+import wiki_index  # noqa: E402
+from wiki_llm import invoke_llm, map_and_validate_citations  # noqa: E402
+
+
+def _load_or_build_index(root):
+    """Reuse the persisted code-index when present; else build in-memory.
+
+    The persisted index lacks chunk text (kept small), so for retrieval we
+    rebuild the full in-memory index. This stays cheap because build_index is
+    dependency-free and capped.
+    """
+    return wiki_index.build_index(root)
+
+
+def _build_prompt(question, chunks):
+    parts = [
+        "Answer the question using ONLY the numbered code excerpts below. "
+        "Cite the excerpts you use with their index in square brackets, like "
+        "[1] or [3]. Cite by index ONLY. Do NOT write file paths yourself. Do "
+        "NOT use emojis. Do NOT use em dashes or en dashes. If the excerpts do "
+        "not contain the answer, say so plainly.",
+        "",
+        "QUESTION: " + question,
+        "",
+        "CODE EXCERPTS:",
+    ]
+    for i, ch in enumerate(chunks, start=1):
+        parts.append("[%d] %s (lines %d-%d):" % (
+            i, ch["file"], ch["start_line"], ch["end_line"]))
+        parts.append(ch["text"])
+        parts.append("")
+    return "\n".join(parts)
+
+
+def _extractive_answer(question, chunks):
+    """Deterministic fallback when no LLM is available.
+
+    Returns prose that references the top chunks by index so the same
+    citation-mapping path applies (every [n] resolves to a real chunk).
+    """
+    lines = [
+        "No language model was available, so here are the most relevant code "
+        "locations for your question:",
+        "",
+    ]
+    for i, ch in enumerate(chunks, start=1):
+        snippet = ch["text"].strip().splitlines()[:3]
+        preview = " ".join(s.strip() for s in snippet)[:160]
+        lines.append("- [%d] %s" % (i, preview))
+    return "\n".join(lines)
+
+
+def main(argv=None):
+    ap = argparse.ArgumentParser(description="Grounded cited codebase Q&A.")
+    ap.add_argument("--root", default=".")
+    ap.add_argument("--question", required=True)
+    ap.add_argument("--k", type=int, default=6)
+    ap.add_argument("--json", action="store_true", help="emit JSON")
+    ap.add_argument("--quiet", action="store_true")
+    args = ap.parse_args(argv)
+
+    root = Path(args.root).resolve()
+    if not root.is_dir():
+        print("error: not a directory: %s" % root, file=sys.stderr)
+        return 2
+
+    index = _load_or_build_index(root)
+    chunks = wiki_index.retrieve(index, args.question, k=max(1, args.k))
+    if not chunks:
+        if args.json:
+            print(json.dumps({"answer": "", "citations": [],
+                              "note": "no relevant code found"}))
+        else:
+            print("No relevant code found for that question.")
+        return 3
+
+    prompt = _build_prompt(args.question, chunks)
+    raw = invoke_llm(prompt)
+    if raw is None or not raw.strip():
+        raw = _extractive_answer(args.question, chunks)
+
+    answer, citations = map_and_validate_citations(raw, chunks, root)
+
+    if args.json:
+        print(json.dumps({
+            "question": args.question,
+            "answer": answer.strip(),
+            "citations": citations,
+        }, indent=2))
+        return 0
+
+    print(answer.strip())
+    if citations:
+        print("")
+        print("Citations:")
+        for c in citations:
+            print("  %s:%d" % (c["file"], c["line"]))
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

package/autonomy/lib/wiki-generator.py

@@ -0,0 +1,322 @@
+#!/usr/bin/env python3
+"""wiki-generator.py -- build a per-project cited wiki under .loki/wiki/ (R5).
+
+Loki's answer to Devin DeepWiki: an architecture overview, key-modules guide,
+and data-flow section generated from the codebase, where EVERY section cites the
+real source files it was built from.
+
+Grounding contract: section citations are CODE-DERIVED. The scanner picks the
+real files and the definition-extractor reads the real def/class/function line
+numbers from those files. The LLM (or template fallback) writes prose only; it
+never invents a citation. Citations are validated against the filesystem before
+being written, so a citation can never point at a file/line that does not exist.
+
+Incremental: a signature over (git HEAD + per-file content hashes) is stored in
+wiki-manifest.json. Re-running skips regeneration when the signature is unchanged
+(unless --force). Same cheap-incremental idea as the proof/docs manifests.
+
+CI-safe: LOKI_WIKI_LLM_STUB mocks the LLM (see wiki_llm.py); with no provider and
+no stub, a deterministic template wiki is written (citations still real).
+
+Invoked as a subprocess (hyphen in filename, like proof-generator.py):
+  python3 wiki-generator.py --root <project> [--force] [--quiet]
+Exit codes: 0 generated or up-to-date, 2 usage/error.
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import os
+import sys
+from datetime import datetime, timezone
+from pathlib import Path
+
+_HERE = os.path.dirname(os.path.abspath(__file__))
+if _HERE not in sys.path:
+    sys.path.insert(0, _HERE)
+
+import wiki_index  # noqa: E402
+from wiki_llm import invoke_llm  # noqa: E402
+
+# Optional reuse of the proof redactor for a final PII sweep. Repo-relative
+# paths already avoid /Users/<name>, but this is belt-and-suspenders.
+try:
+    import proof_redact  # noqa: E402
+    _HAVE_REDACT = True
+except Exception:  # pragma: no cover - redactor optional
+    _HAVE_REDACT = False
+
+
+def _utc_now():
+    return datetime.now(timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ")
+
+
+def _top_modules(index, root, limit=10):
+    """Pick the most significant source files as 'key modules'.
+
+    Significance = chunk count (proxy for size/centrality). Returns a list of
+    {file, defs:[{name,line}], line_count} with REAL definition citations.
+    """
+    counts = {}
+    for ch in index["chunks"]:
+        counts[ch["file"]] = counts.get(ch["file"], 0) + 1
+    ranked = sorted(counts.items(), key=lambda kv: (-kv[1], kv[0]))[:limit]
+    modules = []
+    for rel, n in ranked:
+        defs = wiki_index.extract_definitions(root, rel, limit=8)
+        modules.append({
+            "file": rel,
+            "approx_lines": n * wiki_index.CHUNK_LINES,
+            "defs": defs,
+        })
+    return modules
+
+
+def _entry_points(root, files):
+    """Detect likely entry-point files from the scanned set (real files only)."""
+    candidates = [
+        "src/index.ts", "src/index.js", "src/main.ts", "src/main.js",
+        "index.ts", "index.js", "main.ts", "main.js", "server.ts", "server.js",
+        "src/app.ts", "src/app.js", "app.py", "main.py", "manage.py",
+        "src/main.py", "__main__.py", "main.go", "cmd/main.go",
+        "src/main.rs", "src/lib.rs", "bin/loki",
+    ]
+    fileset = set(files)
+    return [c for c in candidates if c in fileset]
+
+
+def _llm_prose(section, context, fallback):
+    """Get prose for a section from the LLM, or use the deterministic fallback."""
+    prompt = (
+        "You are documenting a software project for a team wiki. Write a clear, "
+        "factual " + section + " section in markdown. Be concise (under 250 words). "
+        "Do NOT use emojis. Do NOT use em dashes or en dashes. Do NOT invent file "
+        "paths. Output ONLY the markdown prose, no headings, no citations.\n\n"
+        "PROJECT CONTEXT:\n" + context
+    )
+    out = invoke_llm(prompt)
+    if out and out.strip():
+        return out.strip()
+    return fallback
+
+
+def _build_context(root, index, modules, entries):
+    name = Path(root).name
+    lines = [
+        "PROJECT: %s" % name,
+        "SOURCE FILES INDEXED: %d" % len(index["files"]),
+        "ENTRY POINTS: %s" % (", ".join(entries) or "none detected"),
+        "TOP MODULES:",
+    ]
+    for m in modules[:8]:
+        lines.append("  - %s (~%d lines)" % (m["file"], m["approx_lines"]))
+    # Top-level dirs.
+    dirs = {}
+    for f in index["files"]:
+        top = f.split("/")[0] if "/" in f else "(root)"
+        dirs[top] = dirs.get(top, 0) + 1
+    lines.append("DIRECTORIES:")
+    for d, n in sorted(dirs.items(), key=lambda kv: -kv[1])[:12]:
+        lines.append("  - %s/ (%d files)" % (d, n))
+    return "\n".join(lines)
+
+
+def _section_overview(root, index, modules, entries, context):
+    name = Path(root).name
+    fallback = (
+        "%s is a software project with %d indexed source files. "
+        "Entry points: %s. The largest modules by size are listed below."
+        % (name, len(index["files"]), ", ".join(entries) or "not detected")
+    )
+    prose = _llm_prose("architecture overview", context, fallback)
+    # Code-derived citations: the entry points + top modules (real files).
+    citations = []
+    for e in entries[:3]:
+        citations.append({"file": e, "line": 1})
+    for m in modules[:4]:
+        line = m["defs"][0]["line"] if m["defs"] else 1
+        citations.append({"file": m["file"], "line": line})
+    return {"id": "architecture", "title": "Architecture Overview",
+            "body": prose, "citations": citations}
+
+
+def _section_modules(root, modules, context):
+    fallback_lines = ["The following modules carry most of the codebase:"]
+    citations = []
+    body_parts = []
+    prose = _llm_prose(
+        "key modules summary",
+        context,
+        "Key modules of this project, with their primary definitions:",
+    )
+    body_parts.append(prose)
+    body_parts.append("")
+    for m in modules:
+        defs = m["defs"]
+        def_str = ", ".join("%s (L%d)" % (d["name"], d["line"]) for d in defs[:6])
+        body_parts.append(
+            "- **%s** (~%d lines)%s"
+            % (m["file"], m["approx_lines"], (": " + def_str) if def_str else "")
+        )
+        line = defs[0]["line"] if defs else 1
+        citations.append({"file": m["file"], "line": line})
+    return {"id": "modules", "title": "Key Modules",
+            "body": "\n".join(body_parts), "citations": citations}
+
+
+def _section_data_flow(root, index, entries, context):
+    fallback = (
+        "Execution begins at the entry point(s) (%s) and flows through the "
+        "key modules. Trace a request from the entry file into the modules it "
+        "imports to follow the data path." % (", ".join(entries) or "not detected")
+    )
+    prose = _llm_prose("data flow", context, fallback)
+    citations = [{"file": e, "line": 1} for e in entries[:4]]
+    if not citations and index["files"]:
+        citations = [{"file": index["files"][0], "line": 1}]
+    return {"id": "data-flow", "title": "Data Flow",
+            "body": prose, "citations": citations}
+
+
+def _validate_citations(root, citations):
+    """Drop any citation that does not resolve to a real file/line on disk."""
+    root = Path(root)
+    clean = []
+    for c in citations:
+        rel = c.get("file")
+        line = int(c.get("line", 1))
+        if not rel:
+            continue
+        abs_path = root / rel
+        try:
+            if not abs_path.is_file():
+                continue
+            with open(abs_path, "r", encoding="utf-8", errors="replace") as f:
+                nlines = sum(1 for _ in f)
+        except OSError:
+            continue
+        if 1 <= line <= max(nlines, 1):
+            clean.append({"file": rel, "line": line})
+    return clean
+
+
+def _render_md(section):
+    lines = ["## %s" % section["title"], "", section["body"], ""]
+    if section["citations"]:
+        lines.append("**Sources:**")
+        for c in section["citations"]:
+            lines.append("- `%s:%d`" % (c["file"], c["line"]))
+        lines.append("")
+    return "\n".join(lines)
+
+
+def main(argv=None):
+    ap = argparse.ArgumentParser(description="Generate a cited project wiki.")
+    ap.add_argument("--root", default=".", help="project root")
+    ap.add_argument("--force", action="store_true", help="regenerate even if unchanged")
+    ap.add_argument("--quiet", action="store_true")
+    args = ap.parse_args(argv)
+
+    root = Path(args.root).resolve()
+    if not root.is_dir():
+        print("error: not a directory: %s" % root, file=sys.stderr)
+        return 2
+
+    def log(msg):
+        if not args.quiet:
+            print(msg)
+
+    wiki_dir = root / ".loki" / "wiki"
+    manifest_path = wiki_dir / "wiki-manifest.json"
+
+    signature = wiki_index.compute_signature(root)
+
+    if manifest_path.is_file() and not args.force:
+        try:
+            prev = json.loads(manifest_path.read_text())
+        except (OSError, json.JSONDecodeError):
+            prev = {}
+        if prev.get("signature") == signature:
+            log("wiki up to date (codebase unchanged); use --force to regenerate")
+            return 0
+
+    log("scanning codebase ...")
+    index = wiki_index.build_index(root)
+    if not index["files"]:
+        print("error: no source files found to document", file=sys.stderr)
+        return 2
+
+    modules = _top_modules(index, root)
+    entries = _entry_points(root, index["files"])
+    context = _build_context(root, index, modules, entries)
+
+    log("generating sections (%d files indexed) ..." % len(index["files"]))
+    sections = [
+        _section_overview(root, index, modules, entries, context),
+        _section_modules(root, modules, context),
+        _section_data_flow(root, index, entries, context),
+    ]
+
+    # Enforce the grounding contract: validate every citation against disk.
+    for sec in sections:
+        sec["citations"] = _validate_citations(root, sec["citations"])
+
+    # Optional PII sweep on prose (paths are already repo-relative).
+    if _HAVE_REDACT:
+        try:
+            proof_redact.set_context(repo_root=str(root))
+            for sec in sections:
+                sec["body"] = proof_redact.redact_value(sec["body"])
+            proof_redact.reset_context()
+        except Exception:
+            pass
+
+    wiki_dir.mkdir(parents=True, exist_ok=True)
+    generated_at = _utc_now()
+    wiki = {
+        "schema_version": "1.0",
+        "project": root.name,
+        "generated_at": generated_at,
+        "file_count": len(index["files"]),
+        "sections": sections,
+    }
+    (wiki_dir / "wiki.json").write_text(json.dumps(wiki, indent=2))
+
+    # Rendered markdown.
+    index_md = ["# %s -- Project Wiki" % root.name, "",
+                "Auto-generated by Loki (R5). %d source files indexed. "
+                "Generated %s." % (len(index["files"]), generated_at), ""]
+    for sec in sections:
+        index_md.append(_render_md(sec))
+        # Per-section file too.
+        (wiki_dir / (sec["id"] + ".md")).write_text(_render_md(sec))
+    (wiki_dir / "index.md").write_text("\n".join(index_md))
+
+    # Persist the code index (used by ask + dashboard).
+    (wiki_dir / "code-index.json").write_text(json.dumps({
+        "root_relative": True,
+        "files": index["files"],
+        "chunks": [
+            {"id": c["id"], "file": c["file"],
+             "start_line": c["start_line"], "end_line": c["end_line"]}
+            for c in index["chunks"]
+        ],
+    }, indent=2))
+
+    manifest = {
+        "schema_version": "1.0",
+        "signature": signature,
+        "generated_at": generated_at,
+        "file_count": len(index["files"]),
+        "sections": [s["id"] for s in sections],
+    }
+    wiki_dir.mkdir(parents=True, exist_ok=True)
+    manifest_path.write_text(json.dumps(manifest, indent=2))
+
+    log("wiki written to %s/.loki/wiki/ (%d sections)" % (root.name, len(sections)))
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())