gabrielmoreira AI Planning And Search Foundations
10 applied AI planning and search primitives for turning a problem into states, actions, constraints, heuristics, and plans. Use this when the hard part is problem formulation or explicit search over alternatives, not language fluency.
Contents
- Quick Reference
- When to Apply
- Primitive Index
- Formal Supporting Theory
- Anti-Patterns
- Misuse Boundaries
- Decision Checklist
- Composition Recipes
- Workflow
- ASCII Flow
- Navigation
- Fact-Checking
Quick Reference
| Primitive | Problem It Solves | Key Parameters |
|---|---|---|
| Problem Formulation | Vague tasks cannot be searched or verified | State, actions, transition model, goal test, path cost |
| Uninformed Search | Need complete baseline without domain heuristic | Branching factor b; depth d; frontier policy |
| Heuristic Search | Large state spaces need directed exploration | Heuristic h(n); admissibility; consistency |
| Local Search | State is large but path is irrelevant | Neighborhood; objective; restart/schedule |
| Constraint Satisfaction | Need assignments satisfying hard constraints | Variables, domains, constraints, MRV/LCV, arc consistency |
| Adversarial Search | Opponent actions affect outcomes | Utility, depth, alpha-beta bounds, rollout policy |
| Classical Planning | Need valid action sequence from symbolic preconditions/effects | STRIPS/PDDL, progression/regression, plan graph |
| Hierarchical Planning | Tasks decompose into reusable subplans | Methods, subtasks, ordering constraints |
| Contingent / Belief-State Planning | Partial observability or nondeterministic actions | Belief state, sensing actions, policy vs sequence |
| Planner-Agent Integration | LLM agent needs explicit plan validity and search boundaries | Planner tool, state abstraction, verifier, replanning trigger |
When to Apply
Apply this skill when:
- The task can be stated as states, actions, transitions, goals, and costs.
- You need A*, uniform-cost search, beam search, backtracking, alpha-beta, MCTS, STRIPS/PDDL, HTN, or CSP reasoning.
- An LLM agent is hallucinating action sequences that need explicit validity checks.
- The workflow needs a planner as a tool, not only prompt decomposition.
- A search or planning algorithm choice changes runtime, completeness, optimality, or failure behavior.
Skip or route elsewhere when:
- It is product search, vector retrieval, ranking, or query matching -> use
software-searchorai-rag. - It is expected utility, value of information, real options, or bandits -> use
foundations-decision-theory. - It is feedback control, setpoint tracking, or MPC -> use
foundations-control-theory. - It is strategic incentive design or equilibrium -> use
foundations-game-theory. - It is cooperative subagent allocation with shared payoff and partitioned information -> use
foundations-team-theory. - It is general agent architecture, memory, tools, or MCP/A2A orchestration -> use
ai-agentsafter this skill defines the planner boundary.
Primitive Index
Each primitive is expanded in references/primitives-overview.md. Use references/patterns-scenarios-traps.md for scenario recipes and references/formal-theory-map.md when completeness, optimality, or complexity claims matter.
| # | Mechanism | Failure Mode Addressed |
|---|---|---|
| 1 | Problem formulation | Agent cannot know what counts as legal progress |
| 2 | Uninformed search | No baseline for completeness, optimality, or frontier explosion |
| 3 | Heuristic search | State space explodes because exploration is undirected |
| 4 | Local search | Path-tracking wastes memory when only final configuration matters |
| 5 | CSP | Constraints are mixed into prompts instead of enforced structurally |
| 6 | Adversarial search | Opponent response is ignored or treated as noise |
| 7 | Classical planning | Preconditions and effects are implicit, so invalid plans pass review |
| 8 | HTN | Repeated task decomposition is ad hoc and inconsistent |
| 9 | Contingent / belief-state planning | Plan assumes full observability or deterministic actions |
| 10 | Planner-agent integration | LLM tool loop lacks plan validation, replanning, or bounded search |
Formal Supporting Theory
| Theory Area | Use When | Applied Primitives It Grounds |
|---|---|---|
| State-space search | Need complete/optimal algorithms over explicit states | #1, #2, #3 |
| Graph algorithms | Need shortest paths, frontier policies, or path-cost proofs | #2, #3 |
| Heuristic admissibility | Need guarantees that A* returns optimal paths | #3 |
| Combinatorial optimization | Need efficient assignment/configuration under hard constraints | #4, #5 |
| Constraint propagation | Need pruning before or during backtracking | #5 |
| Game-tree search | Need bounded lookahead against an opponent | #6 |
| Automated planning | Need symbolic preconditions, effects, and plan validity | #7 |
| Hierarchical task networks | Need reusable decomposition methods | #8 |
| Decision processes under observability limits | Need policies over belief states or contingencies | #9 |
| Neuro-symbolic / tool-using agents | Need LLMs to propose, planners to verify, and executors to act | #10 |
Anti-Patterns
| Anti-Pattern | Planning/Search Diagnosis | Fix |
|---|---|---|
| "Ask the LLM to plan" with no state model | No legal-action or goal-test boundary | Define state, action schema, preconditions, effects, and cost (#1, #7) |
| Greedy search used where optimality is promised | Heuristic is not an admissible cost lower bound | Use A* with admissible/consistent heuristic or stop promising optimality (#3) |
| BFS on high branching factor without depth bound | Frontier blowup | Use UCS/A*, iterative deepening, pruning, or abstraction (#2, #3) |
| Constraints buried in prompt prose | Violations are discovered after execution | Model as CSP with propagation and backtracking (#5) |
| Minimax for real-world negotiation | Payoffs and strategies are not a finite game tree | Route incentive/equilibrium design to foundations-game-theory (#6 boundary) |
| PDDL generated but never validated | Planner accepts malformed or semantically wrong domain | Run domain/problem validation and check plan preconditions/effects (#7) |
| Replanning on every token/tool call | Planner-agent boundary is too fine-grained | Replan only on state drift, failed precondition, or new observation (#10) |
| Treating a valid plan as a safe plan | Validity checks preconditions/effects; it does not check whether the goal or path is one that should be executed | Add a separate safety/permission gate over the action set — validity and safety are independent axes (#7, #10) |
| Adding tree search to raise accuracy without a verifier | Search amplifies the scoring signal; an unreliable scorer just finds higher-confidence errors faster | Establish scorer/verifier reliability first, then spend rollout budget (#3, #6) |
Misuse Boundaries
| Misuse | Why It Is Wrong | Required Correction |
|---|---|---|
| Calling a prompt outline a plan | It has no executable action model or validity check | Convert to state/action/precondition/effect model |
| Treating local optimum as global optimum | Local search is incomplete without strong assumptions | Use restarts, exact search for small instances, or state the approximation |
| Using A* with an overestimating heuristic while claiming optimality | A* optimality depends on admissibility/consistency | Prove h(n) <= true remaining cost or downgrade claim |
| Modeling soft preferences as hard CSP constraints | Search may become infeasible for avoidable reasons | Separate hard constraints from weighted objectives |
| Using adversarial search for cooperative teams | Cooperative information structure has different primitives | Route to foundations-team-theory |
| Treating partial observability as deterministic planning | Actions may need sensing and contingencies | Use belief-state or contingent planning (#9) |
| Inferring plan safety from planning competence | Measured separately, the two do not track each other: a model can be near-perfect at producing executable plans and still route a large fraction of tasks through dangerous actions | Gate the action set independently of the planner's success metric (#10) |
Decision Checklist
- State model: Can you name states, legal actions, transition model, goal test, and cost? -> Problem formulation (#1)
- No heuristic: Need complete baseline over small/medium search? -> BFS/DFS/UCS/IDS (#2)
- Need optimal path with guidance: Can you design an admissible heuristic? -> A* (#3)
- Only final assignment matters: Path is irrelevant and space is huge? -> Local search (#4)
- Hard constraints dominate: Variables/domains/constraints are natural? -> CSP (#5)
- Opponent moves matter: Finite lookahead game/tree? -> Adversarial search (#6)
- Preconditions/effects matter: Need valid action sequence? -> Classical planning (#7)
- Reusable decomposition exists: Tasks break into methods/subtasks? -> HTN (#8)
- Observations are incomplete: Need sensing, contingency, or policy? -> Belief-state planning (#9)
- LLM agent is involved: Need planner/verifier/tool-loop boundary? -> Planner-agent integration (#10)
Composition Recipes
LLM Agent With Valid Plans
Failure: Agent produces plausible steps that violate tool preconditions.
- Problem formulation (#1): define typed state and legal action schema.
- Classical planning (#7): encode preconditions/effects for tools.
- Planner-agent integration (#10): LLM proposes goals or abstractions; planner validates; executor runs only valid next action.
- Replanning trigger (#10): replan after failed precondition, external state drift, or new observation.
Configuration Assistant That Respects Constraints
Failure: Recommendations violate compatibility, license, budget, or availability constraints.
- CSP (#5): model components as variables and compatibility rules as constraints.
- Local search (#4): if preferences are soft and space is too large for exact search.
- Decision theory boundary: route scoring tradeoffs to
foundations-decision-theorywhen expected value or risk weighting matters.
Game Or Simulation Agent
Failure: Agent picks locally good moves and misses opponent responses.
- Adversarial search (#6): minimax or alpha-beta for deterministic games.
- MCTS (#6): stochastic or high-branching games with rollout evaluation.
- Heuristic search (#3): evaluate states with domain heuristic.
- Game theory boundary: route equilibrium, mechanism, or incentive design to
foundations-game-theory.
Workflow Planner For Human Operations
Failure: Task decomposition is repeated manually and varies across runs.
- HTN (#8): encode standard decompositions and method selection.
- Classical planning (#7): validate preconditions for steps with tool or data dependencies.
- Grounding boundary: use
foundations-grounding-communicationfor handoff repair and common-ground checks.
"Works In The Demo, Fails In Production"
Failure: A planner or agent loop that passed every walkthrough degrades or breaks under real traffic.
- Diagnose in order of likelihood before assuming a logic bug: production branching factor vs. demo branching factor, heuristic/verifier fit to the demo distribution, unmodeled nondeterminism, missing or unbounded replanning triggers, a validator that exists in eval but not in the production executor path, and search budget (rollouts/nodes/beam width) silently shrunk by latency or cost limits.
- Full diagnostic checklist:
references/patterns-scenarios-traps.md.
Workflow
- Formulate the problem as state, actions, transition model, goal test, and cost.
- Classify the dominant structure: path search, assignment/CSP, game tree, symbolic plan, hierarchy, or uncertainty.
- Choose the simplest algorithm family that preserves required guarantees: completeness, optimality, bounded memory, or good-enough solution quality.
- Write the search/planning assumptions explicitly before implementation.
- For LLM agents, decide what the LLM may propose and what the planner/verifier must enforce.
- Validate with small hand-checkable cases, impossible cases, and adversarial edge cases.
ASCII Flow
Planning/search problem
-> Define state, actions, transition, goal, cost
-> Classify structure
+-- path to goal -> BFS/UCS/A*/local search
+-- variable assignment -> CSP
+-- opponent -> minimax/alpha-beta/MCTS
+-- symbolic action model -> STRIPS/PDDL/plan graph
+-- reusable decompositions -> HTN
+-- partial observability -> belief-state/contingent planning
-> Check guarantees and constraints
-> Integrate with agent/tool loop if needed
-> Validate plan legality and replanning triggers
Navigation
- Primitives overview:
references/primitives-overview.md - Patterns, scenarios, and traps:
references/patterns-scenarios-traps.md - Formal theory map:
references/formal-theory-map.md - Sources:
data/sources.json
Related Skills
foundations-decision-theory- utility, value of information, regret, and bandits.foundations-control-theory- feedback loops, stability, and MPC.foundations-game-theory- incentives, equilibrium, mechanism design, and strategic actors.foundations-team-theory- cooperative multi-agent allocation with shared payoff.foundations-grounding-communication- handoff repair and common-ground checks for HTN workflow planners.
Fact-Checking
- Check
data/sources.jsonbefore citing algorithm properties or source claims. - Treat complexity and optimality guarantees as conditional on assumptions: nonnegative costs for UCS/A*, admissible/consistent heuristics for A*, finite branching for completeness, and correct action preconditions/effects for planners.
- Do not invent benchmark numbers. If a planner or solver is recommended, report assumptions, problem size, and validation cases instead of generic speed claims.
- LLM planning benchmark numbers move fast and are rarely comparable across papers: success rate depends on domain, instance size, prompt encoding, number of retries, and whether a verifier was in the loop. Cite the specific setup or state the result qualitatively; do not carry a headline percentage across domains.
- Plan validity and plan safety are distinct measurements. A reported planning success rate says nothing about whether the produced plans are safe to execute, and the two have been observed to diverge sharply. Never substitute one metric for the other.
Learnings Loop
Before applying this skill on a non-trivial task, read learnings.consolidated.md in this directory (and learnings.md if present).
After applying it, if you encountered a pattern worth remembering, a mistake worth preventing, or a domain fact that surprised you, append one dated bullet to learnings.md via agents-skills-feedback-loop/scripts/append_learning.py. Do not modify SKILL.md itself.