Documentation
¶
Overview ¶
Package agentassets holds locklog's agent texts (skill, instructions, Cursor rule, example config) and their wiring through lockkit's agentwire.
Index ¶
Constants ¶
This section is empty.
Variables ¶
var ( //go:embed skill.md Skill []byte //go:embed agents.md Instructions []byte //go:embed cursor-rule.mdc CursorFrontMatter []byte //go:embed config.toml ProjectConfig []byte )
Functions ¶
func ClaudeAllow ¶
func ClaudeAllow() []string
ClaudeAllow are the calls Claude Code makes without a prompt: the free locklog tools only, which answer masked data within the budgets or only tighten the masking. An approved tool (a change request; later a scan, an unmask, an export) is never listed: Claude Code asks for it, and the console approves it.
func InitUser ¶
InitUser wires agent for the account (the skill and the user MCP server), never with an allow list: that is written only at project scope, by InitProject.
func Report ¶
func Report(w io.Writer, res InitResult)
Report prints what a wiring wrote, one line per file, the allow entries it added, then its notes and warnings.
Types ¶
type InitResult ¶
type InitResult struct {
agentwire.Result
// AllowAdded are the entries this run added to permissions.allow in
// .claude/settings.local.json, in order.
AllowAdded []string
// Warnings are printed after the notes.
Warnings []string
}
InitResult is what a project init did: agentwire's actions and notes, the allow entries this run added (a re-run adds back the ones the human removed, and says which), and warnings.
func InitProject ¶
func InitProject(cwd, agent string) (InitResult, error)
InitProject wires agent into the project that holds cwd (agentwire's ProjectRoot): `locklog init <agent>`, and what `locklog doctor --fix` runs for each detected agent not wired yet. The allow list is only ever written at project scope, never into the user's settings: a foreign project's server named "locklog" must not get its tools allowed.
Source Files