Agent-native structural language
Sley is built for structured intent, not brittle prompts. Agents and people share the same source surface with deterministic edits and auditable boundaries.
Agent-native structural language
Sley: agent-native structural programming
Sley is a compiler-aware language for practical software editing where bounded structure, verification, and evidence are first-class design constraints.
Auditable claim evidence
The world-first agent-native category claim is treated as a criteria-bound claim. The public repository now carries claim evidence, a machine-readable claim manifest, and a primary-source prior-art packet for independent review.
Proof-first editing
Planned transformations go through bounded checks and receipts first, so edits are structured for handoff, verification, and review across sessions.
May 27 checkpoint
The latest checkpoint keeps the local v1 gate green and expands Sley-owned report-builder registry coverage, runtime source-call paths, FileRead, and deterministic seeded host-effect returns. Strict self-hosting remains a blocked release claim.
May 27 checkpoint / machine facts
Current public status
- Sley is Greyforge Labs' agent-native structural programming language for compiler-mediated, human-reviewed software change.
- The May 27 checkpoint keeps the local v1 gate green while final public release promotion remains blocked.
- The public evidence chain is the Sley home, claim evidence, claim manifest, prior-art source pack, FAQ, and Greyforge Labs context.
- Sensitive host-facing behavior is modeled through deterministic authority gates rather than implicit live provider, shell, network, secret, deployment, payment, or spend actions.
198integration checks
127contract fixtures
23accepted corpus cases
43rejected corpus cases
Language-first agent editing
Plan bounded changes first, then apply only minimal, verified grafts that preserve context and reduce churn.
- Use structured AST, graph slices, and query views to keep edits focused
- Prefer `--dry-run` and JSON receipts before write operations
- Keep edits explainable to both operators and agents
Token efficiency by design
Sley intentionally avoids repeating entire file context every loop.
- Scope checks to one node, task, or symbol slice
- Use receipts and next-actions as compact handoff artifacts
- Avoid copy-heavy edits by using compiler-guided transformations
Public-safe claims posture
Public docs and walkthroughs stay release-safe while surfacing meaningful, reviewable claims.
- Deterministic checks and verifier outputs are machine-readable
- May 27 checkpoint keeps the local v1 gate green
- Open claims link includes release posture and current boundaries
- GitHub and Chronicles provide the canonical project status
structured quickstart
5-step public walkthrough
- 1) Install and inspect your project:
sley doctor --json . - 2) Generate the first structural view:
sley ast --json main.sley - 3) Inspect intent and authority:
sley query --json --kind tasks main.sley - 4) Draft bounded edits:
sley plan --json --graft-templates main.sley - 5) Verify before you promote:
sley verify --json main.sley
citation chain
Auditor packet
Review the public evidence map in Sley Claim Evidence, Sley Claim Manifest, and Sley Prior-Art Source Pack. They define the criteria, local audit commands, current blockers, and official-source comparison posture.
Walkthrough focus
- Keep edits bounded, reviewable, and intention-first.
- Use machine-consumable JSON receipts for downstream handoff and replay.
- Lean on schema-linked diagnostics to keep token churn low.
- Preserve claims boundaries by routing internals through the official repository posture.