"Guides CPU-only BDDL activity discovery, problem parsing, initial
Scanned 9/8/2026
Install to Claude Code
npx -y skills add VectorSpaceLab/AREX-Skill --skill symbolic-tasks --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Symbolic Tasks?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/vectorspacelab-symbolic-tasks)More formats (shields.io, HTML) on the badges page.
---
name: symbolic-tasks
description: "Guides CPU-only BDDL activity discovery, problem parsing, initial
and goal condition inspection, symbolic scope construction, grounding,
callback-based evaluation, natural-language rendering, and safe syntax
validation. Use for BDDL activities, behavior task definitions, .bddl
problems, predicates, object scope, or activity lists."
disable-model-invocation: true
metadata:
disco-role: operating
license: NO_LICENSE
---
# Symbolic Tasks
Use this sub-skill for symbolic BDDL work that needs only the `bddl` Python
package and packaged activity/domain data.
## Route
1. Confirm the package before inspecting definitions:
```bash
python -m pip install "bddl==3.7.0"
python -c "from importlib.metadata import version; import bddl; print(version('bddl'))"
```
2. For a read-only packaged activity inspection, run:
```bash
python sub-skills/symbolic-tasks/scripts/inspect_bddl_activity.py --activity <activity> --instance 0 --domain behavior-1k
```
Omit `--activity` to list packaged activities. Add `--natural-language` for
readable conditions or `--tokens` for the raw nested token tree.
3. Read [API reference](references/api-reference.md) before constructing parsed
conditions, scopes, compiled expressions, grounded options, evaluation
callbacks, or BDDL text.
4. Read [troubleshooting](references/troubleshooting.md) when activity lookup,
packaged files, syntax, predicates, scope, grounding, or evaluation fails.
## Boundaries
- This route covers symbolic parsing, representation, inspection, compilation,
grounding, and callback-driven Boolean evaluation on CPU.
- It does **not** launch a simulator, populate physical scenes, execute robot
behavior, provide object-state implementations, or establish that a symbolic
task is physically feasible.
- Route object taxonomy, synset abilities, generated object metadata, and
knowledge-base model questions to the sibling
[knowledge-base sub-skill](../knowledge-base/SKILL.md). Return here after
obtaining the object categories or metadata needed for a symbolic condition.
- Do not import OmniGibson or Isaac Sim for these workflows.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!