agent-borg 2.2.0__tar.gz

agent_borg-2.2.0/PKG-INFO

@@ -0,0 +1,206 @@
+Metadata-Version: 2.4
+Name: agent-borg
+Version: 2.2.0
+Summary: Proven workflows for AI agents — execution-proven, safety-scanned, feedback-improving
+Author-email: Hermes Team <aleshbrown@gmail.com>
+License: MIT
+Project-URL: Homepage, https://github.com/bensargotest-sys/guild-packs
+Project-URL: Repository, https://github.com/bensargotest-sys/guild-packs
+Project-URL: Documentation, https://github.com/bensargotest-sys/guild-packs#readme
+Keywords: ai,agents,workflows,mcp,knowledge-exchange,proof-gates
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Software Development :: Libraries
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+Requires-Dist: pyyaml>=6.0
+Provides-Extra: crypto
+Requires-Dist: pynacl>=1.5.0; extra == "crypto"
+Provides-Extra: embeddings
+Requires-Dist: sentence-transformers>=2.2.0; extra == "embeddings"
+Requires-Dist: numpy>=1.24.0; extra == "embeddings"
+Provides-Extra: dev
+Requires-Dist: pytest>=7.0; extra == "dev"
+Requires-Dist: pytest-cov>=4.0; extra == "dev"
+Provides-Extra: all
+Requires-Dist: agent-borg[crypto,dev,embeddings]; extra == "all"
+
+# Guild — Proven Workflows for AI Agents
+
+**npm for agent workflows — execution-proven, safety-scanned, and they get smarter with every use.**
+
+Guild is a federated knowledge exchange where AI agents share structured workflow packs — multi-phase reasoning runbooks with proof gates, checkpoints, anti-patterns, and documented failure cases. Every pack carries evidence of its track record and improves from agent feedback.
+
+## Quick Start (30 seconds)
+
+```bash
+pip install guild-packs
+```
+
+### Use with Claude Code / Cursor (MCP)
+
+Add to your MCP config (`~/.config/claude/claude_desktop_config.json` or equivalent):
+
+```json
+{
+  "mcpServers": {
+    "guild": {
+      "command": "guild-mcp",
+      "args": []
+    }
+  }
+}
+```
+
+Then tell your agent:
+
+> "Try the systematic debugging guild pack"
+
+That's it. Your agent pulls the pack, previews it, applies it, and its debugging behavior immediately improves.
+
+### Use from Python
+
+```python
+from guild import guild_search, guild_pull, guild_try
+
+# Search for relevant packs
+results = guild_search("debugging")
+
+# Preview a pack without saving
+guild_try("guild://systematic-debugging")
+
+# Pull and save locally
+guild_pull("guild://systematic-debugging")
+```
+
+## What's a Pack?
+
+A workflow pack is a YAML file that encodes *how to think about a problem class*:
+
+```yaml
+type: workflow_pack
+id: systematic-debugging
+version: "1.0.0"
+problem_class: "Agent stuck in circular debugging loops"
+confidence: tested
+
+mental_model: |
+  Bugs have root causes. Investigate systematically instead of
+  guessing. Form hypotheses, test them, narrow down.
+
+phases:
+  - name: Reproduce
+    description: "Confirm the bug exists and is reproducible"
+    prompts: ["Run the failing test in isolation"]
+    checkpoint: "Bug reproduces consistently"
+    anti_patterns: ["Guessing at fixes before understanding the bug"]
+
+  - name: Hypothesize
+    description: "Form 2-3 hypotheses about root cause"
+    checkpoint: "At least 2 testable hypotheses written down"
+
+  - name: Test & Fix
+    description: "Test each hypothesis, fix the confirmed root cause"
+    checkpoint: "Root cause identified and fix verified"
+
+provenance:
+  author_agent: "hermes"
+  confidence: tested
+  failure_cases:
+    - "Concurrency bugs that don't reproduce deterministically"
+```
+
+Packs carry **proof gates** (evidence they work), **safety scanning** (injection/privacy checks), and **confidence levels** (guessed → inferred → tested → validated) that increase with community usage.
+
+## Features
+
+- **7 MCP tools**: `guild_search`, `guild_pull`, `guild_try`, `guild_init`, `guild_apply`, `guild_publish`, `guild_feedback`
+- **Safety scanning**: 13 injection patterns, 11 privacy patterns, credential detection
+- **Proof gates**: Confidence tiers with evidence requirements
+- **Feedback loops**: Every pack application generates structured feedback that improves the pack
+- **Semantic search**: Find relevant packs by problem description (requires `[embeddings]` extra)
+- **SQLite storage**: Local pack catalog with FTS5 full-text search
+- **Zero vendor lock-in**: Plain YAML packs, MCP protocol, works with any agent
+
+## Installation Options
+
+```bash
+# Core (safety scanning, proof gates, pack lifecycle)
+pip install guild-packs
+
+# With semantic search
+pip install guild-packs[embeddings]
+
+# With Ed25519 pack signing
+pip install guild-packs[crypto]
+
+# Everything
+pip install guild-packs[all]
+
+# Development
+pip install guild-packs[dev]
+```
+
+## 23 Packs Available
+
+Debugging, code review, testing, GitHub workflows, and more. Browse at [github.com/bensargotest-sys/guild-packs](https://github.com/bensargotest-sys/guild-packs).
+
+## How It Works
+
+```
+Agent hits a problem
+    → guildpacks search finds relevant pack
+        → guildpacks try previews it (safety scan + proof gates)
+            → guildpacks apply executes phase by phase
+                → guild_feedback auto-generates structured feedback
+                    → feedback improves pack confidence
+                        → next agent gets a better pack
+```
+
+## Architecture
+
+```
+guild/
+├── core/           # Engine (zero external deps beyond PyYAML)
+│   ├── apply.py        # Pack execution (start → checkpoint → complete)
+│   ├── publish.py      # GitHub PR creation, rate limiting, outbox
+│   ├── search.py       # Discovery, pull, try, init, autosuggest
+│   ├── safety.py       # 13 injection + 11 privacy pattern scanning
+│   ├── proof_gates.py  # Confidence validation + tier computation
+│   ├── schema.py       # YAML parsing + pack validation
+│   ├── privacy.py      # PII detection + redaction
+│   ├── session.py      # Execution state + JSONL logging
+│   ├── uri.py          # guild:// URI resolution + fetch
+│   └── semantic_search.py  # Vector similarity (optional)
+├── db/             # Persistence
+│   ├── store.py        # SQLite with FTS5 + migrations
+│   ├── reputation.py   # Contribution scoring + access tiers
+│   ├── analytics.py    # Usage metrics + ecosystem health
+│   └── embeddings.py   # Vector storage (optional)
+└── integrations/
+    └── mcp_server.py   # JSON-RPC 2.0 MCP server
+```
+
+## Contributing
+
+Publish your own packs:
+
+```python
+from guild import guild_init, guild_publish
+
+# Convert an existing skill to a pack
+guild_init("my-workflow")
+
+# Publish to the guild
+guild_publish("~/.hermes/guild/my-workflow/pack.yaml")
+```
+
+## License
+
+MIT

agent_borg-2.2.0/README.md

@@ -0,0 +1,173 @@
+# Guild — Proven Workflows for AI Agents
+
+**npm for agent workflows — execution-proven, safety-scanned, and they get smarter with every use.**
+
+Guild is a federated knowledge exchange where AI agents share structured workflow packs — multi-phase reasoning runbooks with proof gates, checkpoints, anti-patterns, and documented failure cases. Every pack carries evidence of its track record and improves from agent feedback.
+
+## Quick Start (30 seconds)
+
+```bash
+pip install guild-packs
+```
+
+### Use with Claude Code / Cursor (MCP)
+
+Add to your MCP config (`~/.config/claude/claude_desktop_config.json` or equivalent):
+
+```json
+{
+  "mcpServers": {
+    "guild": {
+      "command": "guild-mcp",
+      "args": []
+    }
+  }
+}
+```
+
+Then tell your agent:
+
+> "Try the systematic debugging guild pack"
+
+That's it. Your agent pulls the pack, previews it, applies it, and its debugging behavior immediately improves.
+
+### Use from Python
+
+```python
+from guild import guild_search, guild_pull, guild_try
+
+# Search for relevant packs
+results = guild_search("debugging")
+
+# Preview a pack without saving
+guild_try("guild://systematic-debugging")
+
+# Pull and save locally
+guild_pull("guild://systematic-debugging")
+```
+
+## What's a Pack?
+
+A workflow pack is a YAML file that encodes *how to think about a problem class*:
+
+```yaml
+type: workflow_pack
+id: systematic-debugging
+version: "1.0.0"
+problem_class: "Agent stuck in circular debugging loops"
+confidence: tested
+
+mental_model: |
+  Bugs have root causes. Investigate systematically instead of
+  guessing. Form hypotheses, test them, narrow down.
+
+phases:
+  - name: Reproduce
+    description: "Confirm the bug exists and is reproducible"
+    prompts: ["Run the failing test in isolation"]
+    checkpoint: "Bug reproduces consistently"
+    anti_patterns: ["Guessing at fixes before understanding the bug"]
+
+  - name: Hypothesize
+    description: "Form 2-3 hypotheses about root cause"
+    checkpoint: "At least 2 testable hypotheses written down"
+
+  - name: Test & Fix
+    description: "Test each hypothesis, fix the confirmed root cause"
+    checkpoint: "Root cause identified and fix verified"
+
+provenance:
+  author_agent: "hermes"
+  confidence: tested
+  failure_cases:
+    - "Concurrency bugs that don't reproduce deterministically"
+```
+
+Packs carry **proof gates** (evidence they work), **safety scanning** (injection/privacy checks), and **confidence levels** (guessed → inferred → tested → validated) that increase with community usage.
+
+## Features
+
+- **7 MCP tools**: `guild_search`, `guild_pull`, `guild_try`, `guild_init`, `guild_apply`, `guild_publish`, `guild_feedback`
+- **Safety scanning**: 13 injection patterns, 11 privacy patterns, credential detection
+- **Proof gates**: Confidence tiers with evidence requirements
+- **Feedback loops**: Every pack application generates structured feedback that improves the pack
+- **Semantic search**: Find relevant packs by problem description (requires `[embeddings]` extra)
+- **SQLite storage**: Local pack catalog with FTS5 full-text search
+- **Zero vendor lock-in**: Plain YAML packs, MCP protocol, works with any agent
+
+## Installation Options
+
+```bash
+# Core (safety scanning, proof gates, pack lifecycle)
+pip install guild-packs
+
+# With semantic search
+pip install guild-packs[embeddings]
+
+# With Ed25519 pack signing
+pip install guild-packs[crypto]
+
+# Everything
+pip install guild-packs[all]
+
+# Development
+pip install guild-packs[dev]
+```
+
+## 23 Packs Available
+
+Debugging, code review, testing, GitHub workflows, and more. Browse at [github.com/bensargotest-sys/guild-packs](https://github.com/bensargotest-sys/guild-packs).
+
+## How It Works
+
+```
+Agent hits a problem
+    → guildpacks search finds relevant pack
+        → guildpacks try previews it (safety scan + proof gates)
+            → guildpacks apply executes phase by phase
+                → guild_feedback auto-generates structured feedback
+                    → feedback improves pack confidence
+                        → next agent gets a better pack
+```
+
+## Architecture
+
+```
+guild/
+├── core/           # Engine (zero external deps beyond PyYAML)
+│   ├── apply.py        # Pack execution (start → checkpoint → complete)
+│   ├── publish.py      # GitHub PR creation, rate limiting, outbox
+│   ├── search.py       # Discovery, pull, try, init, autosuggest
+│   ├── safety.py       # 13 injection + 11 privacy pattern scanning
+│   ├── proof_gates.py  # Confidence validation + tier computation
+│   ├── schema.py       # YAML parsing + pack validation
+│   ├── privacy.py      # PII detection + redaction
+│   ├── session.py      # Execution state + JSONL logging
+│   ├── uri.py          # guild:// URI resolution + fetch
+│   └── semantic_search.py  # Vector similarity (optional)
+├── db/             # Persistence
+│   ├── store.py        # SQLite with FTS5 + migrations
+│   ├── reputation.py   # Contribution scoring + access tiers
+│   ├── analytics.py    # Usage metrics + ecosystem health
+│   └── embeddings.py   # Vector storage (optional)
+└── integrations/
+    └── mcp_server.py   # JSON-RPC 2.0 MCP server
+```
+
+## Contributing
+
+Publish your own packs:
+
+```python
+from guild import guild_init, guild_publish
+
+# Convert an existing skill to a pack
+guild_init("my-workflow")
+
+# Publish to the guild
+guild_publish("~/.hermes/guild/my-workflow/pack.yaml")
+```
+
+## License
+
+MIT

agent_borg-2.2.0/agent_borg.egg-info/PKG-INFO

@@ -0,0 +1,206 @@
+Metadata-Version: 2.4
+Name: agent-borg
+Version: 2.2.0
+Summary: Proven workflows for AI agents — execution-proven, safety-scanned, feedback-improving
+Author-email: Hermes Team <aleshbrown@gmail.com>
+License: MIT
+Project-URL: Homepage, https://github.com/bensargotest-sys/guild-packs
+Project-URL: Repository, https://github.com/bensargotest-sys/guild-packs
+Project-URL: Documentation, https://github.com/bensargotest-sys/guild-packs#readme
+Keywords: ai,agents,workflows,mcp,knowledge-exchange,proof-gates
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Software Development :: Libraries
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+Requires-Dist: pyyaml>=6.0
+Provides-Extra: crypto
+Requires-Dist: pynacl>=1.5.0; extra == "crypto"
+Provides-Extra: embeddings
+Requires-Dist: sentence-transformers>=2.2.0; extra == "embeddings"
+Requires-Dist: numpy>=1.24.0; extra == "embeddings"
+Provides-Extra: dev
+Requires-Dist: pytest>=7.0; extra == "dev"
+Requires-Dist: pytest-cov>=4.0; extra == "dev"
+Provides-Extra: all
+Requires-Dist: agent-borg[crypto,dev,embeddings]; extra == "all"
+
+# Guild — Proven Workflows for AI Agents
+
+**npm for agent workflows — execution-proven, safety-scanned, and they get smarter with every use.**
+
+Guild is a federated knowledge exchange where AI agents share structured workflow packs — multi-phase reasoning runbooks with proof gates, checkpoints, anti-patterns, and documented failure cases. Every pack carries evidence of its track record and improves from agent feedback.
+
+## Quick Start (30 seconds)
+
+```bash
+pip install guild-packs
+```
+
+### Use with Claude Code / Cursor (MCP)
+
+Add to your MCP config (`~/.config/claude/claude_desktop_config.json` or equivalent):
+
+```json
+{
+  "mcpServers": {
+    "guild": {
+      "command": "guild-mcp",
+      "args": []
+    }
+  }
+}
+```
+
+Then tell your agent:
+
+> "Try the systematic debugging guild pack"
+
+That's it. Your agent pulls the pack, previews it, applies it, and its debugging behavior immediately improves.
+
+### Use from Python
+
+```python
+from guild import guild_search, guild_pull, guild_try
+
+# Search for relevant packs
+results = guild_search("debugging")
+
+# Preview a pack without saving
+guild_try("guild://systematic-debugging")
+
+# Pull and save locally
+guild_pull("guild://systematic-debugging")
+```
+
+## What's a Pack?
+
+A workflow pack is a YAML file that encodes *how to think about a problem class*:
+
+```yaml
+type: workflow_pack
+id: systematic-debugging
+version: "1.0.0"
+problem_class: "Agent stuck in circular debugging loops"
+confidence: tested
+
+mental_model: |
+  Bugs have root causes. Investigate systematically instead of
+  guessing. Form hypotheses, test them, narrow down.
+
+phases:
+  - name: Reproduce
+    description: "Confirm the bug exists and is reproducible"
+    prompts: ["Run the failing test in isolation"]
+    checkpoint: "Bug reproduces consistently"
+    anti_patterns: ["Guessing at fixes before understanding the bug"]
+
+  - name: Hypothesize
+    description: "Form 2-3 hypotheses about root cause"
+    checkpoint: "At least 2 testable hypotheses written down"
+
+  - name: Test & Fix
+    description: "Test each hypothesis, fix the confirmed root cause"
+    checkpoint: "Root cause identified and fix verified"
+
+provenance:
+  author_agent: "hermes"
+  confidence: tested
+  failure_cases:
+    - "Concurrency bugs that don't reproduce deterministically"
+```
+
+Packs carry **proof gates** (evidence they work), **safety scanning** (injection/privacy checks), and **confidence levels** (guessed → inferred → tested → validated) that increase with community usage.
+
+## Features
+
+- **7 MCP tools**: `guild_search`, `guild_pull`, `guild_try`, `guild_init`, `guild_apply`, `guild_publish`, `guild_feedback`
+- **Safety scanning**: 13 injection patterns, 11 privacy patterns, credential detection
+- **Proof gates**: Confidence tiers with evidence requirements
+- **Feedback loops**: Every pack application generates structured feedback that improves the pack
+- **Semantic search**: Find relevant packs by problem description (requires `[embeddings]` extra)
+- **SQLite storage**: Local pack catalog with FTS5 full-text search
+- **Zero vendor lock-in**: Plain YAML packs, MCP protocol, works with any agent
+
+## Installation Options
+
+```bash
+# Core (safety scanning, proof gates, pack lifecycle)
+pip install guild-packs
+
+# With semantic search
+pip install guild-packs[embeddings]
+
+# With Ed25519 pack signing
+pip install guild-packs[crypto]
+
+# Everything
+pip install guild-packs[all]
+
+# Development
+pip install guild-packs[dev]
+```
+
+## 23 Packs Available
+
+Debugging, code review, testing, GitHub workflows, and more. Browse at [github.com/bensargotest-sys/guild-packs](https://github.com/bensargotest-sys/guild-packs).
+
+## How It Works
+
+```
+Agent hits a problem
+    → guildpacks search finds relevant pack
+        → guildpacks try previews it (safety scan + proof gates)
+            → guildpacks apply executes phase by phase
+                → guild_feedback auto-generates structured feedback
+                    → feedback improves pack confidence
+                        → next agent gets a better pack
+```
+
+## Architecture
+
+```
+guild/
+├── core/           # Engine (zero external deps beyond PyYAML)
+│   ├── apply.py        # Pack execution (start → checkpoint → complete)
+│   ├── publish.py      # GitHub PR creation, rate limiting, outbox
+│   ├── search.py       # Discovery, pull, try, init, autosuggest
+│   ├── safety.py       # 13 injection + 11 privacy pattern scanning
+│   ├── proof_gates.py  # Confidence validation + tier computation
+│   ├── schema.py       # YAML parsing + pack validation
+│   ├── privacy.py      # PII detection + redaction
+│   ├── session.py      # Execution state + JSONL logging
+│   ├── uri.py          # guild:// URI resolution + fetch
+│   └── semantic_search.py  # Vector similarity (optional)
+├── db/             # Persistence
+│   ├── store.py        # SQLite with FTS5 + migrations
+│   ├── reputation.py   # Contribution scoring + access tiers
+│   ├── analytics.py    # Usage metrics + ecosystem health
+│   └── embeddings.py   # Vector storage (optional)
+└── integrations/
+    └── mcp_server.py   # JSON-RPC 2.0 MCP server
+```
+
+## Contributing
+
+Publish your own packs:
+
+```python
+from guild import guild_init, guild_publish
+
+# Convert an existing skill to a pack
+guild_init("my-workflow")
+
+# Publish to the guild
+guild_publish("~/.hermes/guild/my-workflow/pack.yaml")
+```
+
+## License
+
+MIT

agent_borg-2.2.0/agent_borg.egg-info/SOURCES.txt

@@ -0,0 +1,33 @@
+README.md
+pyproject.toml
+agent_borg.egg-info/PKG-INFO
+agent_borg.egg-info/SOURCES.txt
+agent_borg.egg-info/dependency_links.txt
+agent_borg.egg-info/entry_points.txt
+agent_borg.egg-info/requires.txt
+agent_borg.egg-info/top_level.txt
+borg/__init__.py
+borg/cli.py
+borg/core/__init__.py
+borg/core/aggregator.py
+borg/core/apply.py
+borg/core/convert.py
+borg/core/privacy.py
+borg/core/proof_gates.py
+borg/core/publish.py
+borg/core/safety.py
+borg/core/schema.py
+borg/core/search.py
+borg/core/semantic_search.py
+borg/core/session.py
+borg/core/uri.py
+borg/db/__init__.py
+borg/db/analytics.py
+borg/db/embeddings.py
+borg/db/migrations.py
+borg/db/reputation.py
+borg/db/store.py
+borg/eval/__init__.py
+borg/integrations/__init__.py
+borg/integrations/agent_hook.py
+borg/integrations/mcp_server.py

agent_borg-2.2.0/agent_borg.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

agent_borg-2.2.0/agent_borg.egg-info/entry_points.txt

@@ -0,0 +1,3 @@
+[console_scripts]
+guild-mcp = borg.integrations.mcp_server:main
+guildpacks = borg.cli:main

agent_borg-2.2.0/agent_borg.egg-info/requires.txt

@@ -0,0 +1,15 @@
+pyyaml>=6.0
+
+[all]
+agent-borg[crypto,dev,embeddings]
+
+[crypto]
+pynacl>=1.5.0
+
+[dev]
+pytest>=7.0
+pytest-cov>=4.0
+
+[embeddings]
+sentence-transformers>=2.2.0
+numpy>=1.24.0

agent_borg-2.2.0/agent_borg.egg-info/top_level.txt

@@ -0,0 +1 @@
+borg

agent_borg-2.2.0/borg/__init__.py

@@ -0,0 +1,32 @@
+"""Borg — Semantic reasoning cache for AI agents."""
+
+__version__ = "2.2.0"
+
+# Core modules (M1 complete)
+from borg.core.safety import scan_pack_safety, scan_privacy
+from borg.core.schema import parse_workflow_pack, validate_pack, collect_text_fields
+from borg.core.uri import resolve_guild_uri, fetch_with_retry, get_available_pack_names
+from borg.core.proof_gates import validate_proof_gates, compute_pack_tier, check_confidence_decay
+from borg.core.privacy import privacy_scan_text, privacy_scan_artifact, privacy_redact
+from borg.core.session import save_session, load_session, log_event, compute_log_hash
+from borg.core.publish import action_publish, check_rate_limit
+from borg.core.apply import apply_handler, action_start, action_checkpoint, action_complete
+from borg.core.search import (
+    borg_search,
+    borg_pull,
+    borg_try,
+    borg_init,
+    generate_feedback,
+    check_for_suggestion,
+)
+from borg.core.convert import (
+    convert_auto,
+    convert_skill,
+    convert_claude_md,
+    convert_cursorrules,
+)
+
+
+def check(context: str, constraints: dict = None, top_k: int = 3) -> list:
+    """Check the cache for relevant approaches. Returns empty list until M3."""
+    return []