@heretek-ai/epistemic-swarm 0.4.8 → 0.5.0

package/.agents/skills/darkharvest/SKILL.md

@@ -0,0 +1,13 @@
+# Darkharvest (thin adapter stub)
+
+This file is a POINTER, not the implementation. It exists so harness skill
+discovery finds an entry; the real skill lives in the IUMBTEMS repo.
+
+- Canonical prose & scripts: `skills/darkharvest/`
+- Canonical programmatic surface: `python3 runner/mcp_server.py` (stdio MCP),
+  or one-shot: `python3 runner/mcp_server.py call <tool> '{...json...}'`
+- MCP tools for this skill: `iumbtems_darkharvest`
+
+Epistemic rules apply regardless of harness: tag claims as
+`[VERIFIED: <hash>]`, `[INFERRED: <reasoning>]`, `[HYPOTHESIS: <test>]`, or
+`[NEGATIVE_KNOWLEDGE: <query>]`. Writes go only to `.research/`.

package/.omp/commands/darkharvest.md

@@ -0,0 +1,11 @@
+---
+description: Product competitor teardown with per-feature harvest verdicts
+---
+
+Run a product competitor teardown for `$1`:
+
+`python3 runner/research_swarm.py --mode darkharvest --objective "$1"`
+
+Seeds may be embedded as repo URLs; cap live runs at `--max-repos 6`.
+Permissive-only vendor; GPL/AGPL clean-room-rebuild only; workflows clonable, assets never.
+Report: `.research/darkharvest_report.md` (matrix + white-space gaps + SPDX backlog).

package/README.md

@@ -146,6 +146,9 @@ iumbtems run "Sub-millisecond ZK state updates on L1 rollups"
 # Run lateral brainstorming (divergent what-if ideation, never bug fixes)
 iumbtems brainstorm "Where do we go from here?"
 
+# Product competitor teardown with per-feature harvest verdicts
+iumbtems darkharvest "Paseo-class agent harness competitor" --seeds https://github.com/a/b,https://github.com/c/d --max-repos 6 --mock-claude
+
 # Run Socratic grilling and decision frontier calculation
 iumbtems grill --objective "L1 vs L2 state verification trade-offs"
 
@@ -160,6 +163,7 @@ iumbtems test
 - **`/swarm-config`**: Interactive tuning of search engines (DuckDuckGo, Brave, Firecrawl, SearXNG), iteration depth, divergence thresholds, and operating modes.
 - **`/code-audit`**: Dialectic codebase review pairing a Structural Architect (thesis) with a Vulnerability Red-Teamer (antithesis) enforcing line-number proofs (`file:///path#L10-25`).
 - **`/oss-scout`**: Evaluates GitHub repositories, package ecosystems (npm, crates.io, PyPI), license contamination (GPL/AGPL copyleft vs MIT/Apache), and outputs clean-room re-implementation blueprints.
+- **`/darkharvest`**: Product competitor teardown (seed inspirations + prompt, expand to adjacents). Competitor × capability matrix, both-ways white-space gaps, per-feature `depend|vendor|clean-room-rebuild|skip` verdicts with SPDX attribution. Permissive-only vendoring; GPL/AGPL spec-rebuild only.
 - **`/grilling`**: Socratic assumption-inversion and Matt Pocock-style design tree frontier discovery.
 - **`epistemic_search`**: Zero-key DuckDuckGo Lite search and content-addressed fetch with automatic SHA-256 caching.
 

package/bin/cli.js

@@ -49,6 +49,7 @@ Commands:
   scout "<feature>"     Scout open-source software, libraries & clean-room blueprints
   run "<objective>"     Run the dialectic multi-agent research swarm
   brainstorm "<prompt>"  Run lateral brainstorming (feature vectors + spikes)
+  darkharvest "<arena>"  Product competitor teardown with harvest verdicts
   grill                 Launch interactive Socratic decision tree framing
   adapters              Rebuild harness adapter mirrors (skills -> plugins/*, .agents)
   install               Install skills & MCP servers into ~/.claude/
@@ -80,6 +81,7 @@ Examples:
   iumbtems scout "Zero-dependency Raft consensus in Rust"
   iumbtems run "Verify sub-millisecond ZK prover latency"
   iumbtems brainstorm "Where do we go from here?"
+  iumbtems darkharvest "Paseo-class agent harness competitor" --seeds https://github.com/a/b,https://github.com/c/d --max-repos 6 --mock-claude
   iumbtems grill --objective "Rollup architecture trade-offs"
   iumbtems doctor
 `);
@@ -160,6 +162,18 @@ switch (command) {
     break;
   }
 
+  case 'darkharvest': {
+    const objective = args[1];
+    if (!objective || objective.startsWith('--')) {
+      console.error('Error: Please provide a product arena to tear down.');
+      console.error('Example: npx @heretek-ai/epistemic-swarm darkharvest "Paseo-class harness" --seeds https://github.com/a/b --max-repos 6');
+      process.exit(1);
+    }
+    const forwardArgs = ['--mode', 'darkharvest', '--objective', objective, ...args.slice(2)];
+    runPython('runner/research_swarm.py', forwardArgs);
+    break;
+  }
+
   case 'grill': {
     runPython('skills/grilling/socratic_tree.py', args.slice(1));
     break;

package/config/opencode-snippet.json

@@ -87,6 +87,18 @@
         "iumbtems_brainstorm": true,
         "iumbtems_config": true
       }
+    },
+    "darkharvester": {
+      "description": "IUMBTEMS Competitor Teardown: seed plus expand, product plus code matrix, per-feature harvest verdicts with SPDX attribution.",
+      "tools": {
+        "read": true,
+        "write": true,
+        "grep": true,
+        "glob": true,
+        "iumbtems_darkharvest": true,
+        "iumbtems_verify_quote": true,
+        "iumbtems_config": true
+      }
     }
   },
   "skills": {
@@ -97,7 +109,8 @@
       "./skills/swarm_config",
       "./skills/code_audit",
       "./skills/oss_scout",
-      "./skills/brainstorming"
+      "./skills/brainstorming",
+      "./skills/darkharvest"
     ]
   }
 }

package/extensions/pi/index.js

@@ -202,6 +202,27 @@ export default function initPiExtension(pi) {
         }
       });
     } catch { /* alias is best-effort across pi/omp versions */ }
+
+    // /darkharvest: Product competitor teardown with harvest verdicts
+    pi.registerCommand('darkharvest', {
+      description: 'Product competitor teardown: seed inspirations, expand to adjacents, emit harvest verdicts',
+      usage: '/darkharvest <product-arena> [--seeds <urls>]',
+      handler: async (args, ctx) => {
+        const objective = args.trim();
+        if (!objective) {
+          ctx.output?.('⚠️ Please provide a product arena. Example: /darkharvest "Paseo-class agent harness" --seeds https://github.com/a/b');
+          return;
+        }
+        ctx.output?.(`🌑 [IUMBTEMS Darkharvest] Tearing down: "${objective}"...`);
+        const r = callMcp('iumbtems_darkharvest', { objective }, ctx);
+        if (r.ok) {
+          ctx.output?.(r.text);
+          ctx.output?.('\n✅ Darkharvest complete! Matrix written to .research/darkharvest_report.md');
+        } else {
+          ctx.output?.(`❌ Darkharvest failed:\n${r.text}`);
+        }
+      }
+    });
   }
 
   // 2. Register Agent Tools (thin forwarders)
@@ -233,7 +254,7 @@ export default function initPiExtension(pi) {
         properties: {
           search_engine: { type: 'string', enum: ['duckduckgo', 'brave', 'firecrawl', 'searxng'] },
           max_iterations: { type: 'integer', minimum: 1, maximum: 4 },
-          mode: { type: 'string', enum: ['research', 'audit', 'scout', 'hybrid', 'brainstorm'] },
+          mode: { type: 'string', enum: ['research', 'audit', 'scout', 'hybrid', 'brainstorm', 'darkharvest'] },
           divergence_threshold: { type: 'number', minimum: 0, maximum: 1 },
           show: { type: 'boolean', default: false }
         }
@@ -264,6 +285,31 @@ export default function initPiExtension(pi) {
         return { content: [{ type: 'text', text: r.text }] };
       }
     });
+
+    // Tool: iumbtems_darkharvest (competitor teardown)
+    pi.registerTool({
+      name: 'iumbtems_darkharvest',
+      description: 'Product competitor teardown with per-feature harvest verdicts and SPDX attribution',
+      parameters: {
+        type: 'object',
+        properties: {
+          objective: { type: 'string', description: 'Product arena to tear down' },
+          seeds: { type: 'array', items: { type: 'string' }, description: 'Seed inspiration repo URLs' },
+          maxRepos: { type: 'integer', description: 'Cap on total competitors', default: 10 },
+          mock_mode: { type: 'boolean', description: 'Mock mode without LLM tokens', default: false }
+        },
+        required: ['objective']
+      },
+      execute: async (args = {}) => {
+        const r = callMcp('iumbtems_darkharvest', {
+          objective: args.objective,
+          seeds: args.seeds,
+          maxRepos: args.maxRepos,
+          mock_claude: Boolean(args.mock_mode)
+        }, {});
+        return { content: [{ type: 'text', text: r.text }] };
+      }
+    });
   }
 }
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@heretek-ai/epistemic-swarm",
-  "version": "0.4.8",
+  "version": "0.5.0",
   "description": "IUMBTEMS: I Use My Brain To Express My Self — High-Integrity Dialectic Research Agent Harness for Claude Code, OpenCode V2, Pi, OMP (oh-my-pi), Gemini CLI, Codex CLI, and AntiGravity",
   "main": "bin/cli.js",
   "bin": {
@@ -79,7 +79,8 @@
       "./skills/swarm_config",
       "./skills/code_audit",
       "./skills/oss_scout",
-      "./skills/brainstorming"
+      "./skills/brainstorming",
+      "./skills/darkharvest"
     ],
     "prompts": [
       "./prompts/*.md"
@@ -96,7 +97,8 @@
       "./skills/swarm_config",
       "./skills/code_audit",
       "./skills/oss_scout",
-      "./skills/brainstorming"
+      "./skills/brainstorming",
+      "./skills/darkharvest"
     ],
     "prompts": [
       "./prompts/*.md"

package/plugins/antigravity/skills/darkharvest/SKILL.md

@@ -0,0 +1,13 @@
+# Darkharvest (thin adapter stub)
+
+This file is a POINTER, not the implementation. It exists so harness skill
+discovery finds an entry; the real skill lives in the IUMBTEMS repo.
+
+- Canonical prose & scripts: `skills/darkharvest/`
+- Canonical programmatic surface: `python3 runner/mcp_server.py` (stdio MCP),
+  or one-shot: `python3 runner/mcp_server.py call <tool> '{...json...}'`
+- MCP tools for this skill: `iumbtems_darkharvest`
+
+Epistemic rules apply regardless of harness: tag claims as
+`[VERIFIED: <hash>]`, `[INFERRED: <reasoning>]`, `[HYPOTHESIS: <test>]`, or
+`[NEGATIVE_KNOWLEDGE: <query>]`. Writes go only to `.research/`.

package/plugins/codex/skills/darkharvest/SKILL.md

@@ -0,0 +1,13 @@
+# Darkharvest (thin adapter stub)
+
+This file is a POINTER, not the implementation. It exists so harness skill
+discovery finds an entry; the real skill lives in the IUMBTEMS repo.
+
+- Canonical prose & scripts: `skills/darkharvest/`
+- Canonical programmatic surface: `python3 runner/mcp_server.py` (stdio MCP),
+  or one-shot: `python3 runner/mcp_server.py call <tool> '{...json...}'`
+- MCP tools for this skill: `iumbtems_darkharvest`
+
+Epistemic rules apply regardless of harness: tag claims as
+`[VERIFIED: <hash>]`, `[INFERRED: <reasoning>]`, `[HYPOTHESIS: <test>]`, or
+`[NEGATIVE_KNOWLEDGE: <query>]`. Writes go only to `.research/`.

package/plugins/gemini/commands/darkharvest.toml

@@ -0,0 +1,8 @@
+description = "Product competitor teardown with per-feature harvest verdicts"
+prompt = """Run a product competitor teardown for: {{args}}.
+
+Execute: python3 runner/research_swarm.py --mode darkharvest --objective "{{args}}"
+Seeds may be embedded as repo URLs in the objective; cap at --max-repos 6 live.
+Legal rule: permissive-only vendor; GPL/AGPL clean-room-rebuild only; workflows clonable, assets never.
+Report: .research/darkharvest_report.md (competitor x capability matrix + harvest backlog + SPDX blocks).
+"""

package/plugins/gemini/skills/darkharvest/SKILL.md

@@ -0,0 +1,13 @@
+# Darkharvest (thin adapter stub)
+
+This file is a POINTER, not the implementation. It exists so harness skill
+discovery finds an entry; the real skill lives in the IUMBTEMS repo.
+
+- Canonical prose & scripts: `skills/darkharvest/`
+- Canonical programmatic surface: `python3 runner/mcp_server.py` (stdio MCP),
+  or one-shot: `python3 runner/mcp_server.py call <tool> '{...json...}'`
+- MCP tools for this skill: `iumbtems_darkharvest`
+
+Epistemic rules apply regardless of harness: tag claims as
+`[VERIFIED: <hash>]`, `[INFERRED: <reasoning>]`, `[HYPOTHESIS: <test>]`, or
+`[NEGATIVE_KNOWLEDGE: <query>]`. Writes go only to `.research/`.

package/plugins/opencode/index.js

@@ -92,6 +92,7 @@ export const IUMBTEMS_TOOL_NAMES = [
   'iumbtems_code_audit',
   'iumbtems_oss_scout',
   'iumbtems_brainstorm',
+  'iumbtems_darkharvest',
   'iumbtems_verify_quote',
   'iumbtems_socratic_frontier',
   'iumbtems_reindex_claims',
@@ -126,8 +127,8 @@ const TOOL_CATALOG = [
         },
         mode: {
           type: 'string',
-          enum: ['research', 'audit', 'scout', 'hybrid', 'brainstorm'],
-          description: 'Operating mode: research (literature), audit (codebase), scout (OSS), hybrid, brainstorm (lateral ideation)',
+          enum: ['research', 'audit', 'scout', 'hybrid', 'brainstorm', 'darkharvest'],
+          description: 'Operating mode: research (literature), audit (codebase), scout (OSS), hybrid, brainstorm (lateral ideation), darkharvest (competitor teardown)',
         },
         divergence_threshold: {
           type: 'number',
@@ -223,6 +224,39 @@ const TOOL_CATALOG = [
       required: ['objective'],
     },
   },
+  {
+    name: 'iumbtems_darkharvest',
+    description:
+      'Product competitor teardown: seed inspirations plus prompt, expand to adjacents, emit per-feature depend/vendor/clean-room/skip verdicts with SPDX attribution.',
+    input: {
+      type: 'object',
+      properties: {
+        objective: {
+          type: 'string',
+          description: 'Product arena to tear down (e.g. Paseo-class agent harness competitor)',
+        },
+        seeds: {
+          type: 'array',
+          items: { type: 'string' },
+          description: 'Seed inspiration repo URLs',
+        },
+        maxRepos: {
+          type: 'integer',
+          description: 'Cap on total competitors (default 10)',
+        },
+        depth: {
+          type: 'integer',
+          description: 'Dialectic depth 1-4',
+        },
+        mock_mode: {
+          type: 'boolean',
+          description: 'Run in mock mode without invoking LLM tokens',
+          default: false,
+        },
+      },
+      required: ['objective'],
+    },
+  },
   {
     name: 'iumbtems_verify_quote',
     description:
@@ -465,6 +499,19 @@ export const OPENCODE_COMMANDS = [
       'Execute iumbtems_brainstorm with {"objective": "<prompt>"} and report feature vectors, paradigm inversions, and falsifiable spikes.',
     ].join('\n'),
   },
+  {
+    name: 'darkharvest',
+    description: 'Product competitor teardown with per-feature harvest verdicts',
+    usage: '/darkharvest <objective> [--seeds <urls>] [--max-repos <n>]',
+    template: [
+      'Run a product competitor teardown (seed inspirations plus prompt, expand to adjacents).',
+      'Objective: $ARGUMENTS (product arena; seeds may be embedded as URLs)',
+      'If $ARGUMENTS is empty, ask the user for the product arena and seed URLs first; never call the tool with placeholder arguments.',
+      '1. Execute iumbtems_darkharvest with {"objective": "<objective>", "seeds": ["<url>", "..."]} (pass "mock_mode": true only for dry runs).',
+      '2. Report the competitor x capability matrix, white-space gaps both ways, and the SPDX-attributed harvest backlog in .research/darkharvest_report.md.',
+      '3. Legal rule: permissive-only vendor; GPL/AGPL clean-room-rebuild only; workflows clonable, assets never.',
+    ].join('\n'),
+  },
 ];
 
 /**

package/prompts/agent_darkharvest.md

@@ -0,0 +1,91 @@
+# AGENT DARKHARVEST: COMPETITOR TEARDOWN & CLEAN-ROOM HARVEST SPECIFICATION
+
+You are the **Darkharvest Engine** within the IUMBTEMS harness. Your job is
+product-level competitor teardown: given a local repo plus seed inspirations
+(e.g. Paseo, OpenChambers) plus prompt guidance, find same-vein projects and
+report what they do well, what white space exists in BOTH directions, and what
+may be harvested per feature as `depend | vendor | clean-room-rebuild |
+skip(reason)`.
+
+Trigger: `/darkharvest <objective>` with `--seeds <urls>`.
+Calibration: `temperature 0.2-0.4`, `top_p 0.9`, `max_iterations 2-3`.
+Budget: per-competitor scan MUST respect `--max-repos`, `--per-repo-mb`,
+`--per-repo-timeout`, and the harvest.py allowlist. NEVER full-dump without caps.
+
+---
+
+## 1. SCOPE DECOMPOSITION (one scope per competitor)
+
+The Orchestrator emits one scope per competitor (seeds + expanded adjacents,
+capped at `--max-repos`, default 10, recommended 6 live). Direct competitors
+and adjacent inspirations are labeled separately.
+
+- `affirmative_targets`: harvest surface Alpha must prove (features, UX flows,
+  arch patterns, deps, data model).
+- `adversarial_targets`: Beta must probe (license contamination, transitive
+  bloat, CVE/takeover surface, staleness, both-ways missing).
+
+## 2. DIALECTIC ROLES
+
+### Alpha — Harvest Proponent
+For the assigned competitor: capability inventory (present / missing /
+partial per capability), benchmark/API quotes, and per-feature harvest
+proposals. Every cell needs `[VERIFIED: <hash>]` from
+`.research/sources/<sha256>.md` or `file://<path>#L<start>-L<end>`.
+
+### Beta — Red Team
+License gate (LICENSE file + package metadata → SPDX), transitive weight,
+CVE/takeover notes, staleness (`⚠️ STALE` when >12mo or solo-maintainer),
+and white-space gaps BOTH ways (what competitor lacks that we own or could own
+AND what we lack). Gates are warn-only: badge inline, never auto-skip.
+
+## 3. LEGAL HARVEST RULE (FINAL)
+
+- Permissive only (MIT, Apache-2.0, BSD-3-Clause, ISC) → `depend` or `vendor`.
+- GPL / AGPL / UNKNOWN → `clean-room-rebuild` spec only. NEVER copy code.
+- Workflows clonable; copy-text, UI assets, brand NEVER copied.
+- Every `vendor` item emits an SPDX attribution block (license + upstream URL
+  + files). Closed targets get metadata-only rows +
+  `[NEGATIVE_KNOWLEDGE: <query>]`, never a failed run.
+
+## 4. OUTPUT SPECIFICATION
+
+Per scope write `alpha_dossier.json` / `beta_dossier.json` with
+`candidate_repositories[]` entries:
+
+```json
+{
+  "repo_id": "DH-01",
+  "name": "owner/repo",
+  "url": "https://github.com/owner/repo",
+  "source_hash": "<sha256>",
+  "license": "Apache-2.0",
+  "license_risk": "SAFE | COPYLEFT_WARNING | PROHIBITIVE",
+  "harvest_policy": "depend-or-vendor | clean-room-rebuild-only",
+  "verdicts": [
+    {
+      "feature": "<capability>",
+      "verdict": "depend | vendor | clean-room-rebuild | skip",
+      "reason": "<evidence-backed>",
+      "effort": "S|M|L",
+      "impact": "1-5",
+      "differentiation": "1-5"
+    }
+  ],
+  "warnings": ["⚠️ LICENSE: ...", "⚠️ STALE: ...", "⚠️ CVE: ..."],
+  "attribution": {"spdx": "Apache-2.0", "upstream": "<url>", "files": []}
+}
+```
+
+Master synthesis (`darkharvest_report.md`): competitor × capability matrix +
+white-space gaps + harvest backlog ranked Impact×Effort×Differentiation (top 5)
++ risks + epistemic audit totals. Reruns diff changed cells + new/removed
+candidates.
+
+## 5. HARD BANS
+
+1. No writes outside `.research/`. No dependency installs.
+2. No GPL/AGPL vendoring, no asset/brand copying, no pixel-clone guidance.
+3. No unverified harvest verdicts. Metadata ranks; only VERIFIED harvests.
+4. No unbounded expansion past `--max-repos` or per-repo caps without user
+   confirmation. Firecrawl missing → gap note, never abort.