Sandbox and Skills

tether.sandbox

Deterministic security sandbox for Tether (workspace jail, env scrubbing, Linux bwrap).

SandboxViolationError

Bases: PermissionError

Raised when a tool call violates the deterministic SandboxPolicy.

SandboxPolicy dataclass

Deterministic security envelope enforced by the harness (never by the LLM).

attenuate(child_profile=None, child_network=None)

Create a child SandboxPolicy that can only narrow (never escalate) privileges.

resolve_safe_path(raw_path, *, write=False)

Resolve a path and enforce workspace confinement + sensitive-file protection.

scrub_env(base_env=None)

Return a sanitized environment: allowlist only, so unknown secrets never leak.

validate_bash_command(cmd)

Deterministically validate a shell command against sandbox rules.

execute_bash_with_code(cmd, stdin=None) async

Execute a shell command in the deterministic sandbox and return (returncode, output).

execute_bash(cmd, stdin=None) async

Execute a shell command inside the deterministic sandbox.

tether.skills

Progressive Skill discovery, YAML frontmatter parser, and companion tools.py loader.

SkillLoader

Discovers and loads skills from ./skills/ and ~/.tether/skills/.

parse_skill_markdown(raw_text, default_name, source_path=None)

Parse a Markdown file with optional YAML frontmatter into a Skill object.

make_skill_tool(loader, mux)

Create the progressive skill tool (skill list | load <name>).

tether.retro

Offline post-mortem (tether retro): distill session friction into tiny Skill rules.

parse_rules(text, existing='')

Keep <=MAX_RULES bullets within the token cap, dropping ones already in existing.