Convert gitgram/ and aitimemachine/ from embedded git repos (which showed as opaque submodule gitlinks on GitHub) into fully-tracked first-party content. AITM now merges its 7 vendored repos as first-party engines: - memory: beads (Go task DAG) + context7 (docs SDK), unified into saved_contexts - git: vendored C git (bundled binary) + gitnexus (in-process code intelligence) - agentic: a2a (@a2a-js/sdk) + pal (AITM-managed Python subprocess) Includes the full session's product build: Gitgram web/mobile/backend features, AITM website/CLI/MCP, the unified memory layer, engine build tooling (scripts/build-engines.sh, Dockerfile, CI), docs (README, ENGINES.md, ARCHITECTURE.md), and extensive test suites. Build artifacts, node_modules, external/, and gemini-cli are gitignored. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Bash Agent Example
A bash script demonstrating how an AI agent can use bd to manage tasks autonomously.
Features
- Pure bash implementation (no Python/Node required)
- Colorized terminal output
- Automatic work discovery
- Random issue creation to simulate real agent behavior
- Dependency linking with
discovered-from - Statistics display
Prerequisites
- bash 4.0+
- bd installed:
curl -fsSL https://raw.githubusercontent.com/gastownhall/beads/main/scripts/install.sh | bash - jq for JSON parsing:
brew install jq(macOS) orapt install jq(Linux) - A beads database initialized:
bd init
Usage
# Make executable
chmod +x agent.sh
# Run with default 10 iterations
./agent.sh
# Run with custom iteration limit
./agent.sh 20
What It Does
The agent runs in a loop:
- Looks for ready work (no blockers)
- Claims the task (sets status to
in_progress) - "Works" on it (simulates 1 second of work)
- 50% chance to discover a follow-up issue
- If discovered, creates and links the new issue
- Completes the original task
- Shows statistics and repeats
Example Output
🚀 Beads Agent starting...
Max iterations: 10
═══════════════════════════════════════════════════
Beads Statistics
═══════════════════════════════════════════════════
Open: 5 In Progress: 0 Closed: 2
═══════════════════════════════════════════════════
Iteration 1/10
═══════════════════════════════════════════════════
ℹ Looking for ready work...
ℹ Claiming task: bd-3
✓ Task claimed
ℹ Working on: Fix authentication bug (bd-3)
Priority: 1
⚠ Discovered issue while working!
✓ Created issue: bd-8
✓ Linked bd-8 ← discovered-from ← bd-3
ℹ Completing task: bd-3
✓ Task completed: bd-3
Use Cases
Continuous Integration
# Run agent in CI to process testing tasks
./agent.sh 5
Cron Jobs
# Run agent every hour
0 * * * * cd /path/to/project && /path/to/agent.sh 3
One-off Task Processing
# Process exactly one task and exit
./agent.sh 1
Customization
Edit the script to customize behavior:
# Change discovery probability (line ~80)
if [[ $((RANDOM % 2)) -eq 0 ]]; then # 50% chance
# Change to:
if [[ $((RANDOM % 10)) -lt 3 ]]; then # 30% chance
# Add assignee filtering
bd ready --json --assignee "bot" --limit 1
# Add priority filtering
bd ready --json --priority 1 --limit 1
# Add custom labels
bd create "New task" -l "automated,agent-discovered"
Integration with Real Agents
This script is a starting point. To integrate with a real LLM:
- Replace
do_work()with calls to your LLM API - Parse the LLM's response for tasks to create
- Use issue IDs to maintain context
- Track conversation state in issue metadata
See Also
- ../python-agent/ - Python version with more flexibility
- ../git-hooks/ - Automatic export/import on git operations