Lambdas

Deterministic scripts that run without an LLM, used in lambda and cron pipeline steps.

Lambdas are reusable script definitions stored in the database and executed on the worker. They run deterministically without an LLM, making them suitable for tasks like running tests, checking git status, calling external APIs, or performing file operations.

Data model

Field	Type	Description
`id`	UUID	Unique identifier
`name`	string	Human-readable name
`description`	string?	Optional description
`path`	string	Relative path within `~/.nenjo/lambdas/`
`body`	string	The script source code
`interpreter`	string	How to execute: `auto`, `bash`, `sh`, `python3`, or `node`
`is_system`	boolean	Whether this is a built-in system lambda
`created_by`	UUID?	Owner user ID (null for system lambdas)

Interpreters

Value	Description
`auto`	Execute the script directly (must have a shebang line and be executable)
`bash`	Run with `/bin/bash`
`sh`	Run with `/bin/sh`
`python3`	Run with `python3`
`node`	Run with `node`

The default interpreter is auto.

Script storage

Lambda scripts are stored in two places:

Database -- The body field contains the script source code. This is the source of truth for the API.
Disk -- Scripts are written to ~/.nenjo/lambdas/{path} on the worker filesystem. The path field determines the file location relative to the lambdas directory.

Environment variables

When a lambda runs as part of a pipeline, the executor injects NENJO_* environment variables from the PipelineContext:

Variable	Source
`NENJO_PIPELINE_ID`	Pipeline ID
`NENJO_STEP_ID`	Current step ID
`NENJO_STEP_NAME`	Current step name
`NENJO_PROJECT_ID`	Project ID
`NENJO_EXECUTION_RUN_ID`	Execution run ID
`NENJO_TICKET_ID`	Ticket ID
`NENJO_TICKET_TITLE`	Ticket title
`NENJO_TICKET_SLUG`	URL-safe ticket identifier
`NENJO_GIT_BRANCH`	Current working branch
`NENJO_GIT_TARGET_BRANCH`	Target branch for merges/PRs
`NENJO_GIT_WORK_DIR`	Worktree directory path
`NENJO_GIT_REPO_URL`	Remote repository URL
`NENJO_CONTEXT_FILE`	Path to a JSON file with the full pipeline context
`NENJO_GIT_COMMON_DIR`	Path to the main `.git/` directory (handles worktrees)

Context file

A JSON file is written to a temporary path and its location is provided via NENJO_CONTEXT_FILE. This file contains the full pipeline context for scripts that need structured data beyond what environment variables provide.

Exit codes

The script's exit code determines the step result:

Exit code	Result
`0`	`passed: true`
Non-zero	`passed: false`

The script's stdout is captured as the step's output text.

Step configuration

When using a lambda in a pipeline step:

{
  "name": "Run Tests",
  "step_type": "lambda",
  "lambda_id": "LAMBDA_UUID",
  "config": {
    "interpreter_override": "bash",
    "timeout": "10m"
  }
}

Config field	Type	Default	Description
`lambda_id`	UUID	--	Can also be set as a top-level step field
`interpreter_override`	string	--	Override the lambda's default interpreter
`timeout`	string	`"5m"`	Maximum execution time

API

Lambda routes are under /api/v1/lambdas and require authentication.

List lambdas

curl -H "Authorization: Bearer $TOKEN" \
  https://your-instance.com/api/v1/lambdas

Create a lambda

curl -X POST \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Run Tests",
    "description": "Execute the project test suite",
    "path": "scripts/run-tests.sh",
    "body": "#!/bin/bash\nset -e\nnpm test",
    "interpreter": "bash"
  }' \
  https://your-instance.com/api/v1/lambdas

Get a lambda

curl -H "Authorization: Bearer $TOKEN" \
  https://your-instance.com/api/v1/lambdas/$LAMBDA_ID

Update a lambda

curl -X PATCH \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "body": "#!/bin/bash\nset -e\nnpm run test:ci"
  }' \
  https://your-instance.com/api/v1/lambdas/$LAMBDA_ID

Delete a lambda

curl -X DELETE \
  -H "Authorization: Bearer $TOKEN" \
  https://your-instance.com/api/v1/lambdas/$LAMBDA_ID

#!/bin/bash
# Lambda: check-git-status.sh
# Fails if there are uncommitted changes

cd "$NENJO_GIT_WORK_DIR" || exit 1

if [ -n "$(git status --porcelain)" ]; then
  echo "Uncommitted changes detected:"
  git status --short
  exit 1
fi

echo "Working directory is clean"
exit 0

Example: PR merge check (cron lambda)

#!/bin/bash
# Lambda: check-pr-merged.sh
# For use in a cron step -- polls until a PR is merged

cd "$NENJO_GIT_WORK_DIR" || exit 1

# Read PR number from context file
PR_NUMBER=$(cat "$NENJO_CONTEXT_FILE" | python3 -c "import sys,json; print(json.load(sys.stdin).get('pr_number',''))")

if [ -z "$PR_NUMBER" ]; then
  echo '{"cron_status": "abort", "reason": "No PR number in context"}'
  exit 0
fi

STATE=$(gh pr view "$PR_NUMBER" --json state -q '.state' 2>/dev/null)

case "$STATE" in
  MERGED)
    echo "{\"cron_status\": \"pass\", \"reason\": \"PR #$PR_NUMBER merged\"}"
    ;;
  CLOSED)
    echo "{\"cron_status\": \"fail\", \"reason\": \"PR #$PR_NUMBER closed without merge\"}"
    ;;
  *)
    echo "{\"cron_status\": \"wait\"}"
    ;;
esac