What is activate_skill()?
activate_skill transitions a skill from discover_skills manifest metadata into live instructions where markdown bodies procedural checklists and bundled references fold into the agent prompt stack. Companion tools or knowledge hooks described by the skill become reachable without loading every skill at cold start which keeps baseline prompts small while specialists opt into heavy playbooks only when triggers match.
Call activation at decision points you can justify such as after before_kickoff inspects inputs when a router tags a branch as compliance heavy or immediately before a task whose expected output depends on a niche domain pack. Avoid unconditional activation in every task because you pay tokens and widen tool surfaces for models that may misuse rarely needed capabilities.
Operational teams should version SKILL markdown like code because breaking changes to headings or tool names ripple through activation order. Pair activation logs with crewai releases so you can diff which skill pack shipped with a regression. Security reviewers should treat third party skills like dependency updates with provenance checks before regulated deployments.
Testing activation paths caches discover_skills output once per process then asserts unknown names raise before hitting production routers. Staging environments should measure prompt token deltas per activation so finance teams can forecast costs when skills grow. Document fallback behavior when activation fails mid kickoff so agents degrade gracefully instead of halting entire crews. Add structured logging around activation boundaries with tenant identifiers so multi tenant hosts can audit which skill bodies entered each kickoff without dumping full markdown into logs.
Use Cases
- • Just-in-time skill loading
- • Cost-aware prompt sizing
Key Features
- ✓ Lazy injection
- ✓ Composable with tools and knowledge
When NOT to Use
Always-on skills — pre-load them via crew config.
Notes
Name drift vs filesystem
Manifest names come from SKILL metadata. Typos or case sensitivity issues surface as runtime errors — validate against discover_skills output before calling activate in production routers.
Token accounting
Large SKILL bodies can dominate the context window. Monitor usage after activation and split oversized skills into smaller focused packs.
Tool duplication
If a skill bundles tools already attached to the agent, you may double-register capabilities. Prefer declarative YAML or explicit tool lists that reconcile with skill metadata.
Security review
Skills are code-adjacent content. Treat third-party SKILL like dependency updates — review before distributing to regulated environments.
Parameters
| Parameter | Type | Required | Purpose |
|---|---|---|---|
| name | str | No | Skill name (from manifest). |
Code Examples
Use
activate_skill('crewai_flows')Conditional activation from inputs
from crewai.skills import activate_skill
def maybe_activate_compliance(inputs: dict) -> None:
if inputs.get('regulated'):
activate_skill('financial_disclosures')Pair with cached discovery
from crewai.skills import activate_skill, discover_skills
SKILLS = {s.name for s in discover_skills()}
def safe_activate(name: str) -> None:
if name not in SKILLS:
raise ValueError(f'Unknown skill {name}')
activate_skill(name)When to Use
When the agent needs a specific skill for the upcoming task.
Common Mistakes
❌ Activating every skill 'just in case'
✅ Activate selectively to control prompt size.
Related: @tool decorator reference, Agent class reference, and the first Crew tutorial.
activate_skill() FAQ
What is activate_skill() in CrewAI?
Loads a skill's SKILL.md body into the agent's instruction context. activate_skill transitions a skill from discover_skills manifest metadata into live instructions where markdown bodies procedural checklists and bundled references fold into the agent prompt stack. Companion tools or knowledge hooks described by the skill become reachable without loading every skill at cold start which keeps baseline prompts small while specialists opt into heavy playbooks only when triggers match. Call activation at decision points you can justify such as a…
Which CrewAI types expose the method activate_skill()?
DevShelfHub documents activate_skill() on Skills. The reference maps it to Python module crewai.skills — pin your installed crewai version and match imports to the snippet on this page.
When should I use activate_skill()?
When the agent needs a specific skill for the upcoming task.
When should I avoid activate_skill()?
Always-on skills — pre-load them via crew config.
How do I call activate_skill() from Python?
activate_skill('my_skill')
Where can I explore more CrewAI API reference pages?
Open the CrewAI API reference index on DevShelfHub to search classes, methods, and decorators, each with runnable examples, parameters, common mistakes, and cross-links.