@polycode-projects/seonix 0.9.1 → 0.10.1

package/README.md

@@ -46,13 +46,13 @@ network calls, **$0**.
 
 ```bash
 # 1. index your repo (deterministic, offline) — writes a machine-local .seonix/
-seonix cli index_repository '{"repo_path":"/abs/path/to/repo"}'
+cd /path/to/repo && seonix cli index_repository
 
 # 2. self-locate the touched modules from the task text (no human-chosen target)
-seonix cli seonix_search '{"repo_path":"/abs/path/to/repo","query":"<your task>"}'
+seonix cli seonix_search '{"query":"<your task>"}'
 
 # 3. render the bounded edit digest for those modules
-seonix cli digest '{"repo_path":"/abs/path/to/repo","modules":["<top hits>"]}' > digest.txt
+seonix cli digest '{"modules":["<top hits>"]}' > digest.txt
 
 # 4. inject digest.txt once at session start, then run your coding agent
 claude --prompt "Use the seonix digest below to localise and make the change. $(cat digest.txt)"
@@ -60,6 +60,13 @@ claude --prompt "Use the seonix digest below to localise and make the change. $(
 
 Add `.seonix/` to your `.gitignore`; it is machine-local and rebuilt from source.
 
+The JSON argument is optional on every `cli` command: an omitted `repo_path`
+resolves to the nearest ancestor of the current directory (current directory
+first) containing a `.seonix/`, so query commands work from any subdirectory of
+an indexed repo. `index_repository` falls back to the current directory when
+nothing is indexed yet; query commands print a one-line error instead. An
+explicit `repo_path` always wins.
+
 ## Indexing multiple repositories
 
 `index_repository` takes one of three argument forms:
@@ -117,14 +124,17 @@ should start it with cwd = the indexed worktree.
 ### `cli index_repository` — build the graph
 
 ```bash
+seonix cli index_repository                       # no arg: index the repo you are in
 seonix cli index_repository '{"repo_path":"/abs/repo"}'
 seonix cli index_repository '{"repo_paths":["/abs/repo1","/abs/repo2"],"out_root":"/abs/dir"}'
 seonix cli index_repository '{"multi_root":"/abs/estate"}'
 ```
 
 `repo_path` / `repo_paths`+`out_root` / `multi_root` are mutually exclusive — pass
-exactly one (or let `seonix.toml`'s `repositories` supply the set when none is
-given explicitly). Shared JSON options:
+at most one. When none is given, `seonix.toml`'s `repositories` supplies the set
+if present; otherwise the target defaults to the nearest ancestor of the current
+directory carrying a `.seonix/` (re-index), else the current directory itself
+(first index). Shared JSON options:
 
 - `ignores` (bool, default `true`) — honour the repo's `.seonixignore`. Pass
   `"ignores":false` to index everything regardless (the bench harness always
@@ -235,7 +245,9 @@ seonix cli seonix_context '{"symbol":"buildContextBundle","repo_path":"/abs/repo
 Any sub-command name that isn't one of the above routes straight to the MCP tool
 dispatcher, so any tool in the server's surface — cold or hot — is invokable from
 Bash with no MCP connection. `repo_path` in the JSON arg selects the target
-graph; every other key is passed through as the tool's own arguments.
+graph; omitted, it defaults to the nearest ancestor of the current directory
+containing a `.seonix/` (a clean one-line error when there is none). Every other
+key is passed through as the tool's own arguments.
 
 ### `hook-augment` — PreToolUse Grep/Glob augmenter
 
@@ -316,7 +328,11 @@ seonix supplies the graph through tmct's provider seam.
 Two opt-in/opt-out environment knobs, orthogonal to everything above:
 `SEONIX_STORE=sqlite` turns on the resident `node:sqlite` store as a derived,
 rebuildable companion to `graph.json` (rebuild it explicitly with `cli
-store_rebuild`; unset, nothing sqlite loads). `SEONIX_GRAPH_FORMAT` controls the
+store_rebuild`; unset, nothing sqlite loads). The variable must be **exported**
+(`export SEONIX_STORE=sqlite`) — a plain shell assignment is invisible to the
+`seonix` child process, and the JSON path runs instead. The index summary's
+`store:` line states which mode actually ran (`store: sqlite (graph.db +
+graph.json)` only when `graph.db` was built). `SEONIX_GRAPH_FORMAT` controls the
 graph's on-disk wire format — the interned v2 form is the shipped default;
 set `SEONIX_GRAPH_FORMAT=1` to opt back out to the legacy v1 form.
 
@@ -341,6 +357,7 @@ template — see `templates/seonix.toml` and
   - `include_text`, `include_structure`, `respect_gitignore`, `markdown_sections`, `vue` — recognized and normalized, but **not yet wired to a consumer** (see "unwired keys" below).
 - **`[tune]`** — the locate/digest scoring knobs: `score_gap_k`, `literal_mention` (shipped default `true`), `demote_non_prod`, `call_adjacency`, `impl_of_interface`, `beam_search` (+ `beam_width`), `embed_rank`, `prose_layers`.
 - **`[tune.expansion]`** — graph-traversal expansion mode used by locate/the viewer's navigation: `strategy` (`none` | `beam` | `spiral` | `ppr`; `none` is the byte-identical shipped default), `nodes` (breadth/token-budget cap), `q` (proportion of most-distinctive neighbours followed per step), `depth` (hop radius).
+- **`[interfaces]`** — HTTP route↔handler edges the AST graph can't see on its own: a `Route` node per endpoint, linked to its handler by a `serves` edge, and (JS/TS + tree-sitter C#) an HTTP client call linked to the route it targets by a `callsHttp` edge. Off by default; turn it on with `enabled = true` or `SEONIX_INTERFACES=1` (env wins). Tunable: `frameworks` (default `["aspnet", "express"]`), `verbs`, `combine_controller_prefix`, `prefix_strip`, `match_threshold`.
 - **`[telemetry]`** — `enabled = true` opts in (see [Telemetry](#telemetry) below).
 
 **Precedence, everywhere:** explicit CLI arg > `seonix.toml` value > shipped
@@ -411,11 +428,15 @@ seonix supplies the graph; tmct owns the conversation. The old opt-in
 
 Python (stdlib `ast`), JS/TS (the TypeScript compiler API, including CommonJS
 `require`/`module.exports`), C# (Roslyn, with a tree-sitter fallback) and Java
-(JavaParser, with a tree-sitter fallback), all merged into the one typed graph. The C# Roslyn helper's source ships in the package under `roslyn/`; build
-it on the consumer side (needs the .NET SDK) if you index C#. The Java JavaParser
-helper is a jar built from the repository's `java/` directory (`mvn package`) and
-is not included in the npm package, so a package install indexes Java with the
-tree-sitter fallback.
+(JavaParser, with a tree-sitter fallback), all merged into the one typed graph.
+The C# Roslyn helper's source ships in the package under `roslyn/`. If the .NET
+SDK is on your PATH, seonix builds it the first time it indexes a C# repo, one
+build, cached after that. No `dotnet` on PATH just means it uses the tree-sitter
+fallback, silently, no error. Run `npm run build:roslyn` yourself first if you
+want the dll ready ahead of time, e.g. in a Docker image build. The Java
+JavaParser helper is a jar built from the repository's `java/` directory (`mvn
+package`) and is not included in the npm package, so a package install indexes
+Java with the tree-sitter fallback.
 
 ## Licence & safety
 

package/bin/cli.mjs

@@ -6,6 +6,12 @@
 //   seonix cli index_repository '{"repo_path":"<abs>"}' → deterministic index
 //     (honours <repo>/.seonixignore — gitignore-subset patterns; pass
 //      `"ignores":false` in the JSON arg to index everything regardless)
+//     The JSON arg is OPTIONAL for every `cli` command: an omitted repo_path defaults
+//     to the nearest ancestor of cwd (cwd first) containing a .seonix/ directory —
+//     so a bare `seonix cli index_repository` indexes the repo you are in, and query
+//     commands run from any subdirectory of an indexed repo. index_repository falls
+//     back to cwd when nothing is indexed yet (it creates the index); query commands
+//     error cleanly instead. An explicit repo_path always wins, unchanged.
 //   seonix cli index_repository '{"repo_paths":["<abs1>","<abs2>",…],"out_root":"<abs-dir>"}' → index
 //     n repos into ONE merged graph at <out_root>/.seonix/ (module ids prefixed with each repo's
 //     directory basename; out_root defaults to the paths' deepest common ancestor directory).
@@ -39,7 +45,7 @@
 // graph artifact to <repo_path>/.seonix/graph.json; the server (started with
 // cwd = that repo) loads it by default. No flags, no config files.
 
-import { join } from "node:path";
+import { dirname, join } from "node:path";
 import { existsSync } from "node:fs";
 import { readFile, writeFile } from "node:fs/promises";
 import { fileURLToPath } from "node:url";
@@ -52,7 +58,7 @@ import * as source from "../src/source.mjs";
 import { parseEntities, renderSearch, rankModulesByProximity, searchModulesRanked, selectRankedModules, DEFAULT_SCORE_GAP } from "../src/codegraph.mjs";
 import { loadTomlConfig, normalizeConfig, mergeEffective, CONFIG_FILE } from "../src/toml-config.mjs";
 import { compileGlobs } from "../src/walk.mjs";
-import { createTelemetry } from "../src/telemetry.mjs";
+import { createTelemetry, truncateStr, looksLikeMiss } from "../src/telemetry.mjs";
 
 /** Build a config pointed at a specific repo's artifact (for `cli` sub-commands that
  *  take a repo_path), or fall back to the cwd-derived default. */
@@ -67,6 +73,39 @@ function parsePayload(payload) {
   catch { return null; }
 }
 
+/** Nearest ancestor of `startDir` (itself first) containing a .seonix/ directory, or null. */
+function findIndexRoot(startDir = process.cwd()) {
+  let dir = startDir;
+  for (;;) {
+    if (existsSync(join(dir, ".seonix"))) return dir;
+    const parent = dirname(dir);
+    if (parent === dir) return null;
+    dir = parent;
+  }
+}
+
+/** Default an omitted repo_path for a `cli` command. Precedence:
+ *    1. explicit repo_path — always wins, args returned untouched (byte-identical to today);
+ *    2. SEONIX_GRAPH_FILE — also untouched (configFor's loadConfig fallback consumes it);
+ *    3. the nearest ancestor of cwd (cwd first) carrying a .seonix/ → written into args.repo_path;
+ *    4. nothing found: commands that READ an index (`mustExist`) get a one-line error + exit 2;
+ *       index_repository falls back to cwd (it CREATES the index there).
+ *  Mutating args means downstream configFor/tomlFor/telemetry all see the resolved repo. */
+function defaultRepoPath(args, { mustExist = true } = {}) {
+  if (args.repo_path !== undefined) return args;
+  // The env override only steers commands that READ a graph (query commands); index_repository
+  // writes to <repo>/.seonix and never consumes SEONIX_GRAPH_FILE, so it still defaults.
+  if (mustExist && process.env.SEONIX_GRAPH_FILE && process.env.SEONIX_GRAPH_FILE.trim()) return args;
+  const root = findIndexRoot();
+  if (root) { args.repo_path = root; return args; }
+  if (mustExist) {
+    process.stderr.write("seonix: no .seonix index found here or above — run `seonix cli index_repository` first or pass repo_path\n");
+    process.exit(2);
+  }
+  args.repo_path = process.cwd();
+  return args;
+}
+
 /** Load a repo's seonix.toml as a normalized config, or null. Returns null IMMEDIATELY when
  *  `args.config === false` (the BENCH-IMMUNITY flag — an arm's index/locate/digest must NEVER read a
  *  subject repo's seonix.toml) OR when no seonix.toml exists. So on the bench path, and with no
@@ -163,6 +202,7 @@ async function runDigest(args) {
   // stdout is the digest injected into the no-MCP arm, so telemetry must never touch it. The
   // seonix.toml [telemetry] toggle rides in via `toml`; SEONIX_TELEMETRY still wins both ways.
   const tel = createTelemetry({ env: process.env, config: configFor(repoPath), toml, surface: "cli" });
+  const t0 = Date.now();
   const tune = resolveDigestTune(args, toml);
   let modules = Array.isArray(args.modules) ? args.modules.slice(0, DIGEST_MODULE_CAP) : [];
   let autoSelected = null; // for the header, when `query` drove selection
@@ -236,10 +276,14 @@ async function runDigest(args) {
   process.stdout.write(out);
   // One telemetry line for the digest surface: the selected modules + the effective tier/topup the
   // header already carries, plus the injected body size. `tier` here MUST equal the header's tier=.
+  // (wh dogfood backlog #6): duration + a miss flag — a query-mode digest that selected zero
+  // modules is an unambiguous miss (autoSelected is only set in query mode).
   tel?.record({
     surface: "digest",
-    query: args.query ? { raw: args.query } : undefined,
+    query: args.query ? { raw: truncateStr(args.query) } : undefined,
     response: { count: emitted, node_ids: modules, tier: effTier, topup },
+    perf: { ms_total: Date.now() - t0 },
+    quality: { miss: autoSelected ? autoSelected.length === 0 : emitted === 0 },
     cost: { returned_chars: out.length, returned_tokens_est: Math.ceil(out.length / 4) },
   });
 }
@@ -320,15 +364,21 @@ async function main() {
       }
       // seonix.toml: config dir = the explicit repo/estate path, else out_root, else cwd (where the
       // file sits). Skipped entirely on the bench path (config:false → tomlFor null → byte-identical).
-      const toml = await tomlFor(args.repo_path || args.multi_root || args.out_root || process.cwd(), args);
+      let toml = await tomlFor(args.repo_path || args.multi_root || args.out_root || process.cwd(), args);
       warnUnwired(toml);
       // repositories: seonix.toml MAY supply the repo set when NO explicit path arg is given
       // (explicit repo_path/repo_paths/multi_root always win).
       const tomlRepos = (given.length === 0 && toml && Array.isArray(toml.repositories) && toml.repositories.length)
         ? toml.repositories : null;
       if (given.length === 0 && !tomlRepos) {
-        process.stderr.write("seonix: index_repository requires repo_path, repo_paths or multi_root\n");
-        process.exit(2);
+        // No path arg and no seonix.toml repo set: default to the nearest ancestor of cwd
+        // carrying a .seonix/ (re-index it), else cwd itself (first index — it creates .seonix/).
+        defaultRepoPath(args, { mustExist: false });
+        // The resolved root may carry its own seonix.toml (cwd can be a subdirectory) — re-read there.
+        if (args.repo_path && args.repo_path !== process.cwd()) {
+          toml = await tomlFor(args.repo_path, args);
+          warnUnwired(toml);
+        }
       }
       // A5 `history_depth`: unified cap over BOTH git-history passes. 0 = skip history entirely;
       // N>0 = `git log -n N`; absent = today's defaults. Precedence: explicit arg > seonix.toml
@@ -449,7 +499,7 @@ async function main() {
         process.stderr.write("seonix: digest expects a JSON arg, e.g. '{\"repo_path\":\"/abs\",\"modules\":[…]}'\n");
         process.exit(2);
       }
-      await runDigest(args);
+      await runDigest(defaultRepoPath(args));
       return;
     }
 
@@ -463,6 +513,24 @@ async function main() {
         process.stderr.write("seonix: seonix_locate expects a JSON arg, e.g. '{\"query\":\"…\",\"repo_path\":\"/abs\"}'\n");
         process.exit(2);
       }
+      // Tool honesty (wh dogfood T7): seonix_locate used to print NOTHING and exit 0 for
+      // any input without a query ({}, {"symbol":…}, {"bogus":1}) — an empty ranking is
+      // indistinguishable from success in a tool loop. Bad/missing args are now a one-line
+      // error naming the valid keys + a nonzero exit, like the generic dispatchTool route.
+      const LOCATE_KEYS = [
+        "query", "raw_query", "repo_path", "config", "limit", "spiral", "spiral_depth",
+        "demote_nonprod", "call_adjacency", "impl_of_interface", "beam_search", "beam_width", "literal_mention",
+      ];
+      const unknownKeys = Object.keys(args).filter((k) => !LOCATE_KEYS.includes(k));
+      if (unknownKeys.length) {
+        process.stderr.write(`seonix: unknown argument(s) for seonix_locate: ${unknownKeys.join(", ")}. Valid: ${LOCATE_KEYS.join(", ")}.\n`);
+        process.exit(2);
+      }
+      if (!String(args.query || "").trim()) {
+        process.stderr.write("seonix: seonix_locate requires a non-empty \"query\" (free text to rank modules against). For a single symbol use seonix_search or seonix_describe.\n");
+        process.exit(2);
+      }
+      defaultRepoPath(args);
       const config = configFor(args.repo_path);
       // seonix.toml [tune] defaults the locate levers (arg > seonix.toml > default). Skipped on the
       // bench path (config:false → tomlFor null) and when no seonix.toml exists → byte-identical.
@@ -471,6 +539,7 @@ async function main() {
       // Opt-in telemetry (default OFF → null → no-op, no file). stdout is the ranked list the rig
       // parses, so telemetry is LOG FILE ONLY. seonix.toml [telemetry] rides in via `toml`.
       const tel = createTelemetry({ env: process.env, config, toml, surface: "cli" });
+      const t0 = Date.now();
       const tune = resolveLocateTune(args, toml);
       try {
         const graph = parseEntities(await source.fetchEntities(config));
@@ -510,15 +579,18 @@ async function main() {
             : {}),
         });
         process.stdout.write(ranked.map((r) => `${r.path}\t${r.score}`).join("\n") + "\n");
-        // Telemetry for the locate surface: the ranked count + the top few paths/scores.
+        // Telemetry for the locate surface: the ranked count + the top few paths/scores, plus
+        // (wh dogfood backlog #6) duration + a miss flag (an empty ranking is an unambiguous miss).
         tel?.record({
           surface: "locate",
-          query: { raw: String(args.query || "") },
+          query: { raw: truncateStr(String(args.query || "")) },
           response: {
             count: ranked.length,
             node_ids: ranked.slice(0, 3).map((r) => r.path),
             scores: ranked.slice(0, 3).map((r) => r.score),
           },
+          perf: { ms_total: Date.now() - t0 },
+          quality: { miss: ranked.length === 0 },
         });
       } catch (e) {
         process.stderr.write(`seonix: ${e?.message || e}\n`);
@@ -539,6 +611,7 @@ async function main() {
         process.stderr.write("seonix: browser_link expects a JSON arg, e.g. '{\"query\":\"classes that change with render\"}'\n");
         process.exit(2);
       }
+      defaultRepoPath(args);
       const config = configFor(args.repo_path);
       const { buildTemporalGraph, gitCommitOrder } = await import("../src/browser.mjs");
       const { nlToQuery, validateLink, encodeViewState, VIEW_DEFAULTS } = await import("../src/temporal.mjs");
@@ -578,6 +651,7 @@ async function main() {
         process.stderr.write("seonix: store_rebuild expects a JSON arg, e.g. '{\"repo_path\":\"/abs\"}'\n");
         process.exit(2);
       }
+      defaultRepoPath(args);
       const { graphFile } = configFor(args.repo_path);
       const { readFile, stat } = await import("node:fs/promises");
       const { writeStore, storeFileFor, verifyStore } = await import("../src/store.mjs");
@@ -612,17 +686,23 @@ async function main() {
         process.stderr.write(`seonix: ${sub} expects a JSON arg, e.g. '{"symbol":"<name>"}'\n`);
         process.exit(2);
       }
+      defaultRepoPath(args);
       const config = configFor(args.repo_path);
       // seonix.toml [telemetry] toggle for the generic tool surface (byte-identical with no toml:
       // tomlFor short-circuits to null on the bench path and when no seonix.toml exists).
       const toml = await tomlFor(args.repo_path, args);
       const tel = createTelemetry({ env: process.env, config, toml, surface: "cli" });
+      const t0 = Date.now();
       try {
         const text = await dispatchTool(sub, args, { config });
         process.stdout.write(text + "\n");
-        // Telemetry for the generic tool surface: which tool + how many chars it returned.
+        // Telemetry for the generic tool surface: which tool + how many chars it returned, plus
+        // (wh dogfood backlog #6) the args text/duration/hit-miss the log previously omitted.
         tel?.record({
           tool: sub,
+          query: { raw: truncateStr(JSON.stringify(args)) },
+          perf: { ms_total: Date.now() - t0 },
+          quality: { miss: looksLikeMiss(text) },
           cost: { returned_chars: text.length, returned_tokens_est: Math.ceil(text.length / 4) },
         });
       } catch (e) {
@@ -632,7 +712,8 @@ async function main() {
       return;
     }
 
-    process.stderr.write("seonix: `cli` needs a sub-command (index_repository | digest | <toolName>)\n");
+    process.stderr.write("seonix: `cli` needs a sub-command (index_repository | digest | <toolName>). " +
+      "The JSON arg is optional — an omitted repo_path defaults to the nearest ancestor of cwd with a .seonix/ index.\n");
     process.exit(2);
   }
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@polycode-projects/seonix",
-  "version": "0.9.1",
+  "version": "0.10.1",
   "private": false,
   "type": "module",
   "description": "SEONIX — a deterministic, offline, $0 code-graph tool that makes a cheap coding agent edit like an expensive one. Indexes a repo with Python ast + git (zero model calls) and renders a bounded edit-digest.",
@@ -35,6 +35,7 @@
     "src/",
     "roslyn/*.cs",
     "roslyn/*.csproj",
+    "scripts/build-roslyn.mjs",
     "README.md",
     "LICENSE"
   ],
@@ -45,19 +46,13 @@
     "infra"
   ],
   "scripts": {
+    "build:roslyn": "node scripts/build-roslyn.mjs",
     "setup": "node scripts/setup.mjs",
     "bench": "node bench/run.mjs",
     "bench:quick": "node bench/run.mjs --instances 1 --runs 1",
     "join:telemetry": "node scripts/join-telemetry.mjs",
     "bench:lite": "node bench/run.mjs --suite lite",
     "bench:seonix": "node bench/run.mjs --suite django-lh --arms seonix,seonix-min --runs 5",
-    "bench:b010": "node bench/run.mjs --suite django-lh --arms otb,seonix,seonix-min,seonix-b010 --models claude-opus-4-8,claude-sonnet-4-6 --runs 5 --concurrency 10",
-    "bench:b011": "node bench/run.mjs --suite django-lh --arms seonix,seonix-min --models claude-opus-4-8,claude-sonnet-4-6,claude-haiku-4-5 --runs 5 --concurrency 15",
-    "bench:b011-haiku-refs": "node bench/run.mjs --suite django-lh --arms otb,seonix-b010 --models claude-haiku-4-5 --runs 5 --concurrency 10",
-    "bench:b015": "node scripts/bench-b015.mjs",
-    "bench:b016p1": "node scripts/bench-b016-p1.mjs",
-    "bench:b016h": "node scripts/bench-b016-h.mjs",
-    "bench:b016p2": "node scripts/bench-b016-p2.mjs",
     "bench:baseline": "node scripts/use-baseline.mjs",
     "proof": "node scripts/proof-demo.mjs",
     "smoke": "node bench/smoke.mjs --suite django-lh",
@@ -92,7 +87,7 @@
   },
   "dependencies": {
     "@modelcontextprotocol/sdk": "^1.29.0",
-    "@polycode-projects/the-mechanical-code-talker": "^0.8.1",
+    "@polycode-projects/the-mechanical-code-talker": "^0.8.2",
     "cytoscape": "^3.30.0",
     "smol-toml": "^1.7.0",
     "tree-sitter": "^0.21.1",

package/scripts/build-roslyn.mjs

@@ -0,0 +1,11 @@
+#!/usr/bin/env node
+// build-roslyn.mjs — explicit prewarm command (`npm run build:roslyn`) for the
+// Roslyn C# semantic extractor. NOT run automatically on `npm install` — the
+// dll lazy-builds on first real C# extraction attempt instead (src/cs_roslyn.mjs
+// via src/roslyn_build.mjs), so consumers who never touch C# never pay a .NET
+// build cost at install time. Use this command ahead of time when you want the
+// dll baked in up front (e.g. a Docker image build step, or a CI cache warm)
+// rather than paying the lazy-build cost on first use.
+import { ensureRoslynBuilt } from "../src/roslyn_build.mjs";
+
+await ensureRoslynBuilt();